SPONSORS

Main Sponsor
Philips Medical Systems

Sponsors
Development Corporation City of Rotterdam (OBR)

Port of Rotterdam Authority

Boston Scientific

OLDelft

Special Grants
Interuniversity Cardiology The Dutch Foundation for
Institute of Ultrasound in
the Netherlands (ICIN) Medicine & Biology (SUGB)

1
2
 TABLE OF CONTENTS

Symposium Organizing Committee .................................. 7

Technical Program Committee ......................................... 8
UFFC AdCom.................................................................. 12
Invited Papers ................................................................. 14
Short Courses ................................................................. 15
Plenary Session .............................................................. 30
Exhibits............................................................................ 31

Session PS: STUDENT PAPER FINALISTS ................. 32

Session P1A: BLOOD FLOW ......................................... 51
Session P1B: ELASTOGRAPHY.................................... 58
Session P1C: TISSUE CHARACTERIZATION .............. 64
Session P1D: BEAMFORMER IMPLEMENTATION ...... 68
Session P1E: MEDICAL IMAGING ................................ 74
Session P1F: ACOUSTIC SENSORS ............................ 83
Session P1G: NDE MODELING
AND MEASUREMENTS ............................................ 90
Session P1H: ULTRASONIC MOTORS/LAB
ON A CHIP ............................................................... 100
Session P1I: PIEZOELECTRICAL
TRANSFORMERS/MOTORS .................................. 106
Session P1J: SIMULATION OF SAW DEVICES ......... 111
Session P1K: SAW DEVICES I .................................... 119
Session P1L: MATERIALS FOR TRANSDUCERS ...... 128
Session P1M: TRANSDUCER MODELING ................. 135
Session 1A: LOCAL DRUG DELIVERY UCA/
THERAPEUTIC ........................................................ 142
Session 2A: STATIC ELASTICITY ............................... 147
Session 3A: BIOSENSORS .......................................... 154
Session 4A: PHONONIC CRYSTALS .......................... 159
Session 5A: BAW THIN FILM FILTERS ....................... 163
Session 6A: HIGH FREQUENCY ARRAYS ................. 169
Session 1B: IVUS I ....................................................... 174
Session 2B: BLOOD FLOW ......................................... 180

3
Session 3B: NDE IMAGING AND SIGNAL
PROCESSING ......................................................... 186
Session 4B: OPTICAL INTERACTIONS ...................... 193
Session 5B: THIN FILM BAW MATERIALS I ............... 197
Session 6B: PIEZOELECTRIC SINGLE CRYSTAL
TRANSDUCERS ...................................................... 202
Session 1C: IVUS II ...................................................... 206
Session 2C: ULTRASOUND 19TH CENTURY
TO 21ST CENTURY ................................................... 212
Session 3C: NDE-WAVE PROPAGATION ................... 217
Session 4C: ULTRASONICS MOTORS ....................... 223
Session 5C: THIN FILM BAW MATERIALS II .............. 230
Session 6C: TRANSDUCERS AND MATERIALS
MODELING ............................................................... 234
Session 1D: MOUSE IMAGING ................................... 241
Session 2D: VASCULAR .............................................. 247
Session 3D: NDE-GENERAL METHODS .................... 254
Session 4D: PHYSICAL ACOUSTICS I........................ 261
Session 5D: NOVEL SAW DEVICES
AND ANALYSIS ....................................................... 265
Session 6D: ULTRASOUND MEMS TECHNOLOGY .. 271
Session P2A: TISSUE CHARACTERIZATION ............ 276
Session P2B: CONTRAST AGENTS I ......................... 283
Session P2C: THERAPEUTICS AND
HYPERTHERMIA .................................................... 289
Session P2D: ELASTICITY IMAGING ......................... 303
Session P2E: NDE SIGNAL PROCESSING
AND MODELING ...................................................... 309
Session P2F: BAW PIEZOELECTRIC FILMS ............. 318
Session P2G: BAW MATERIALS AND
PROPAGATION II .................................................... 324
Session P2H: SAW MATERIALS AND
PROPAGATION ....................................................... 330
Session P2I: SAW DEVICES II .................................... 337
Session P2J: TRANSDUCER MATERIALS
AND COMPOSITE MATERIALS .............................. 343
Session P2K: MICROMACHINED TRANSDUCERS ... 347

4
Session 1E: PORTABLE DEVICES AND OTHER NEAT
STUFF ...................................................................... 358
Session 2E: CARDIOVASCULAR ................................ 363
Session 3E: NDE SIGNAL PROCESSING .................. 369
Session 4E: PHYSICAL ACOUSTICS 2 ....................... 375
Session 5E: RF FILTERS AND MODULES .................. 380
Session 6E: CMUTS ..................................................... 384
Session 1F: ELASTICITY DYNAMIC ........................... 389
Session 2F: CONTRAST AGENTS: IMAGING I .......... 394
Session 3F: GAS-LIQUID SENSORS .......................... 400
Session 4F: MEDICAL TRANSDUCERS I ................... 405
Session 5F: SAW ANALYSIS ....................................... 410
Session 6F: INDUSTRIAL MEASUREMENTS
AND APPLICATIONS................................................ 415
Session 1G: ELASTICITY ............................................ 421
Session 2G: CONTRAST AGENTS: IMAGING II ......... 427
Session 3G: ULTRASONIC FLOWMETERS ............... 433
Session 4G: INTERVENTIONAL ULTRASOUND ........ 440
Session 5G: SAW SYSTEM APPLICATIONS .............. 445
Session 6G: TISSUE CHARACTERIZATION .............. 449
Session 1H: CONTRAST AGENTS: EFFECTS ........... 457
Session 2H: ACOUSTIC PROPERTIES OF CELLS
AND TISSUES .......................................................... 463
Session 3H: ACOUSTICAL IMAGING .......................... 470
Session 4H: BAW RESONATORS ............................... 476
Session 5H: SAW PROPAGATION .............................. 481
Session 6H: ENERGY HARVESTING & NOVEL
TRANSDUCERS ...................................................... 486
Session P3A: BEAMFORMING .................................... 493
Session P3B: CONTRAST AGENTS II ........................ 501
Session P3C: BONE ..................................................... 507
Session P3D: MODELING AND VISUALIZATION ....... 512
Session P3E: SIGNAL PROCESSING ......................... 519
Session P3F: NDE TRANSDUCERS ........................... 526
Session P3G: GENERAL PHYSICAL ACOUSTICS ..... 534
Session P3H: BAW MATERIALS AND
PROPAGATION I ..................................................... 541

5
Session P3I: SAW FILTERS AND DEVICES ............... 547
Session P3J: SAW SYSTEM APPLICATIONS ............ 559
Session P3K: TRANSDUCER FABRICATION
TECHNOLOGY ......................................................... 561
Session P3L: HIGH FREQUENCY TRANSDUCERS .. 564
Session P3M: MEDICAL TRANSDUCERS II ............... 568
Session 1I: CONTRAST AGENTS
FUNDAMENTALS .................................................... 574
Session 2I: ULTRASOUND AND THERAPEUTICS ..... 579
Session 3I: NDE MATERIAL AND DEFECT
CHARACTERIZATION ............................................. 584
Session 4I: FLUIDIC ACTUATION ............................... 589
Session 5I: SAW FILTERS AND TRANSDUCERS ...... 595
Session 6I: TRANSDUCER MATERIALS .................... 600
Session 1J: QUANTITATIVE CARDIAC IMAGING ...... 606
Session 2J: THERAPY MONITORING ......................... 612
Session 3J: MATERIALS CHARACTERIZATION ........ 618
Session 4J: ULTRASONIC MICROFLUIDICS
AND BULK EFFECTS .............................................. 623
Session 5J: BAW RESONATOR DESIGN
AND ANALYSIS ....................................................... 629
Session 6J: BEAMFORMING II .................................... 635
Session 1K: 3D/4D CARDIAC IMAGING ..................... 641
Session 2K: CAVITATION ............................................. 648
Session 3K: ACOUSTIC SENSORS ............................ 654
Session 4K: PHYSICAL ACOUSTICS III ...................... 660
Session 5K: MICROMECHANICAL AND SAW
RESONATORS ......................................................... 666
Session 6K: MEDICAL SIGNAL PROCESSING
AND CONTRAST...................................................... 671
All Author Index ............................................................. 678
De Doelen Floor Plan ................................................... 694
Condensed Program ............... 696 and Inside Back Cover

6
 SYMPOSIUM ORGANIZING COMMITTEE

General Chair
Ton van der Steen
Erasmus University Medical Centre, Rotterdam, The Netherlands
[Link]@[Link]

Technical Program Chair

John Hossack
University of Virginia, Charlottesville, VA, USA
hossack@[Link]

Short Course/Tutorial Chair

Nico de Jong
Erasmus University Medical Centre, Rotterdam, The Netherlands
[Link]@[Link]

Finance Chair
Herman van de Vaart
vdv@[Link]

Publicity Chair
Sorah Rhee
MEGGITT Endevco, San Juan, Capistrano, CA, USA
[Link]@[Link]

Editorial Chair
Marj Yuhas
Industrial Measurement Systems Inc., Aurora, IL USA
myuhas@[Link]

Exhibit/Sponsor Chair
Chris de Korte
Nijmegen University Medical Centre, Nijmegen, The Netherlands
cldekorte@[Link]

Awards Chair
Reinhard Lerch
Friedrich-Alexander-Universitat, Erlangen, Germany
[Link]@[Link]

7
 TECHNICAL PROGRAM COMMITTEE

GROUP 1: Medical Ultrasonics

Vice Chair: Stanislav Emelianov: University of Texas of Austin,
USA
Olivier Basset: INSA, France
Geneviève Berger: Laboratorie Imagerie Parametrique, France
Richard Chiao: Ultrasound Group, Siemens Medical Solutions,
USA
Lawrence A. Crum: University of Washington, USA
Emad Ebbini: University of Minnesota, USA
Helmut Emert: Electrical Engineering Department, Ruhr University,
Germany
David Evans: Department of Medical Physics, Leicester Royal
Infirmary, UK
Kathy Ferrara: Biomedical Engineering, University of California,
USA
Stuart Foster: Department of Medical Physics, Sunnybrook Health
Sciences Centre, Canada
James Greenleaf: Ultrasound Research, Mayo Clinic, USA
John Hossack, Department of Biomedical Engineering, School of
Engineering & Applied Sciences, University of Virginia, USA
Kullervo Hynynen: Brigham and Women’s Hospital, USA
Peter Hoskins: Senior Research Fellow, Medical Physics, The
University of Edinburgh, Scotland
Michael F. Insana: Biomedical Engineering, University of California,
USA
Jorgen Jensen: Department of Information Technology, Technical
University of Denmark
Nico de Jong: Biomedical Engineering, Erasmus University
Rotterdam, The Netherlands
Hiroshi Kanai: Tohoku University, Japan
Jian-yu Lu: The University of Toledo, USA
Leonardo Masotti: Dipartimento di Ingegneria Elettronica, Univer-
sity di Firenze, Italy
James G. Miller: Physics Department, Washington University, USA
Kathy Nightingale: Biomedical Engineering Department, Duke
University, USA
William O’Brien: Director of Bioacoustics Research Lab, Depart-
ment of Electrical & Computer Engineering, University of Illinois,
USA
Helen Routh: Philips Research, USA
Ton van der Steen: Head of Biomedical Engineering, Erasmus
University Rotterdam, The Netherlands
8
 Tom Thomas: Siemens
Kai Thomenius: GE CRD, USA
Peirro Tortoli: Electronic Engineering Department, University of
Florence, Italy
Keith Wear: FDA Center for Devices and Radiological Health, USA

GROUP 2: Sensors, NDE, and Industrial Applications

Vice Chair: Jafar Saniie: Department of Electrical & Computer
Engineering, Illinois Institute of Technology, USA
Robert C. Addison: Rockwell Science Center, USA
Walter Arnold: Fraunhofer Institute for Nondestructive Testing,
Germany
Narendra K. Batra
Eric S. Furgason: School of Electrical & Computer Engineering,
Purdue University, USA
Donna C. Hurley: National Institute of Standards & Technology,
USA
David A. Hutchins: School of Engineering, University of Warwick,
England
Bernhard Jakoby: Institute of Industrial Electronics and Material
Science, Vienna University of Technology, Austria
Lawrence W. Kessler: Sonoscan Inc., USA
Pierre T. Khuri-Yakub: Stanford University, USA
Jun-ishi Kushibike: Department of Electrical Engineering, Gradu-
ate School of Engineering, Tohoku University, Japan
Lawrence C. Lynnworth: Chief Technologist, GE Panametrics,
USA
Roman Gr. Maev: Professor & Director, Center for Imaging Re-
search & Advanced Material Characterization, Department of
Physics, University of Windsor, Canada
Massimo Pappalardo: Lab of Acustoelectronics, Dipartimento di
Ingegneria Elettronica, University di Roma TRE, Italy
Tony Sinclair: Professor of Mechanical & Industrial Engineering,
Department of Mechanical Engineering, University of Toronto,
Canada
Bernhard R. Tittman: Department of Engineering & Mechanics,
Pennsylvania State University, USA
Jiromaru Tsujino: Faculty of Engineering, Kanagawa University,
Japan
Donald E. Yuhas: Industrial Measurement Systems, Inc., USA
John F. Vetelino: Lab for Surface Science & Technology, University
of Maine, USA

9
GROUP 3: Physical Acoustics
Vice Chair: Kenneth Lakin: TFR Technologies, Inc., USA
Art Ballato: Chief Scientist, US Army CECOM RDEC AMSEL-RD-
CS, USA
Mack Brezaeale: Department of Physics, University of Mississippi,
USA
Jan Brown: JB Consulting, USA
David Hecht
Fred Hickernell
Amit Lal: Assistant Professor, School of Electrical & Computer
Engineering, Cornell University, USA
John Larson
Moises Levy: Department of Physics, Naples, Florida, USA
George Mansfield: Institute of Radio Engineering and Electronics,
Russian Academy of Sciences, Russia
Kiyoshi Nakamura: Department of Electrical & Communication
Engineering, Graduate School of Engineering, Tohuku University,
Japan
Valeri Proklov: Institute of Radio Engineering & Electricity, Russia
Edgar Schmidhammer
Susan Schnieder: Department of Electrical & Computer Engineer-
ing, Marquette University, USA
Bikash Sinha, Schlumberger-Doll Research, USA
Yook-Kong Yong: Department of Civil & Environmental Engineer-
ing, Rutgers University, USA
John Vig: US Army CECOM, AMSEL-RD-C2-PT, USA
Smaine Zeroug: Program Manager, Schlumberger-Doll Research,
USA

GROUP 4: Surface Acoustic Waves

Vice Chair: Don Malocha, School of Electrical Engineering &
Computer Science, University of Central Florida, USA
Benjamin Abbott: Sawtek Inc., USA
Ali Bagi-Wadji: Vienna University of Technology, Austria
Kushal Bhattacharjee: Clarisay, USA
Serguey Biryukov: Surfaces and Interfaces Department, Leibniz
Institute for Solid State and Materials Research Dresden (IFW),
Germany
Yasuo Cho: Research Institute of Electrical Communications,
Tohoku University, Japan
Yasuo Ebata: Director of SAW Core Technology, Fujitsu Media
Device Ltd., Japan
Ken-ya Hashimoto: Department of Electronic & Mechanical
Engineering, Chiba University, Japan

10
 Daniel Hauden: CNRS_LPMO, France
Mitsutaka Hikita: Communications Systems Research Department,
Central Research Laboratory, Hitachi, Ltd., Japan
William D. Hunt: School of Electrical & Computer Engineering,
Georgia Institure of Technology, USA
Shen Jen: RF Monolithics Inc., USA
John A. Kosinski: US Army RDE Command, USA
David Morgan: Impulse Consulting, UK
Mauricio Pereira da Cunha: Assistant Professor, Department of
Electrical & Computer Engineering, University of Maine, USA
Viktor Plesski: SAW Design Bureau, Thomson Microsonics,
Switzerland
Bob R. Potter: Vectron International, USA
Arne Ronnekleiv: Division of Physical Electronics, Norwegian
Institute of Technology, Norway
Clemens C. W. Ruppel: EPCOS AG - SAW RD SAM, Germany
Peter Smith: McMaster University, Canada
Robert Weigel: University of Erlangen-Nuremberg, Germany

GROUP 5: Transducers and Transducer Materials

Vice Chair: Scott Smith, GE Corporate Research & Development,
USA
Levent Degertekin: Woodruff School of Mechanical Engineering,
Georgia Institute of Technology, USA
Jean-Francois Gelly: Thomson Microsonics, France
Hal Kunkel: Philips Ultrasound, USA
Reinhard Lerch: Friedrich-Alexander-Universitat Erlangen-
Nurnberg Lehrstuhl fur Sensorik, Germany
Geoff Lockwood: Stirling Hall, Department of Physics, Queen’s
University, Canada
Clyde Oakley: Vice-president of Probe Development, Tetrad
Corporation, USA
Mark E. Schafer: Sonic Tech Inc., USA
K. Kirk Shung: Bioengineering Department, Pennsylvania State
University, USA
Stephen W. Smith: Department of Biomedical Engineering, Duke
University, USA
Wallace A. Smith: Materials Division, Office of Naval Research,
USA
Yasuhito Takeuchi: Professor, Department of Information &
Computer Science, Faculty of Engineering, Kagoshima Univer-
sity, Japan
Roger H. Tancrell: Aimar Technology Corporation, USA
Vasandara Varadan: Pennsylvania State University, USA

11
 Qiming Zhang: Materials Research Lab, Pennsylvania State
University, USA
Yongrae Roh: Kyungpook National University, Korea
Thomas Shrout: Materials Research Lab, Pennsylvania State
University, USA
Christopher Daft: Sensant Corporation, USA

2005 UFFC-S ADMINISTRATIVE COMMITTEE

SOCIETY OFFICERS

PRESIDENT Gerald V. Blessing

Natl. Institute of Standards & Tech (retired)

PRESIDENT-ELECT Art Ballato

U. S. Army RDECOM CERDEC HQ

VP, FERROELECTRICS Susan Trolier-McKinstry

The Pennsylvania State University

VP, FREQUENCY CONTROL Lute Maleki

Jet Propulsion Laboratory

VP, ULTRASONICS Clemens C. Ruppel

EPCOS AG

VP, PUBLICATIONS Donald Yuhas

Industrial Measurement Systems Inc.

SECRETARY-TREASURER Jacqueline H. Hines

J. H. Hines Consulting

12
 ELECTED ADMINISTRATIVE COMMITTEE MEMBERS

2003 - 2005
Thomas R. Shrout, Pennsylvania State University
2003 - 2005
Mathias Fink, Universite Denis Diderot
2003 - 2005
Kullervo Hynynen, Brigham and Women’s
Hospital Harvard Medical School
2003 - 2005
Mike Garvey, Symmetricom Inc.
2004 - 2006
Victor P. Plessky, GVR Trade SA
2004 - 2006
Nava Setter, EPFL Swiss Federal Institute of
Technology in Lausanne
2004 - 2006
Peter M. Smith, McMaster University
2004 - 2006
Daniel S. Stevens, Vectron International
2005 - 2007
Ruyan Guo, Pennsylvania State University
2005 - 2007
Massimo Pappalardo, University of Roma Tre
2005 - 2007
Leonhard M. Reindl, Albert-Ludwigs-University
of Freiburg
2005 - 2007
Mark E. Schafer, Sonic Tech, Inc.

13
 INVITED SPEAKERS
GROUP 1:
Patrick W. Serruys: ‘Biomedical Engineering in the Catheteriza-
tion Laboratory’
Timothy Leighton: ‘Bubble Acoustics: from Seas to Surgeries’
Peter Wells: ‘Lord Rayleigh’
Jaques Souquet: ‘Telemedicine: What is in Place Today? What
are the Challenges for the Future?’
Kathy Ferrara: ‘Targeted Ultrasound Imaging and Drug Delivery’
Ralph Sinkus: ‘Dynamic MR-Elastography as a Non-Invasive
Imaging Modality: In Vivo Application to Breast, Liver, and
Brain’

GROUP 2:
Chris Lowe: ‘Novel Acoustic Biosensors’
K. Yamanaka: ‘Diffraction-Free Propagation of Collimate Surface
Acoustic Wave on a Sphere Applied for Innovative Gas
Sensors’
Ute Rabe: ‘Surface Characterization Using Ultrasonic Vibration
Modes of Atomic Force Microscope Cantilevers’

GROUP 3:
Paul Muralt: ‘Is There a Better Material for Thin Film BAW
Applications than AIN?’
S. V. Krishnaswamy: ‘Piezoelectric/Ferroelectric Films for
Microwave/MEMS Application - Historical Perspective’
E. V. Charnaya: ‘Ionic Conductivity in Acoustic Crystals’

GROUP 4:
Meirion F. Lewis: ‘SAW and Optical Signal Processing’
Peter Wright: ‘Trends in Integrated Front-End Modules for
Cellular Handsets’

GROUP 5:
Christopher Daft: ‘cMUTs and Electronics for 2-D and 3-D
Imaging: Monolithic Integration, In-Handle Chip Sets and
System Implications’
Stephen M. Fry: ‘A Glimpse into the Future of Intravascular
Ultrasound Technology’
Tim Proulx: ‘Advances in Catheter-Based Ultrasound Imaging’
Jie Chen: ‘Commercialization of Piezoelectric Single Crystals for
Medical Imaging Applications’

14
 SHORT COURSES

All courses are Sunday, September 18.

1 - Medical Ultrasound Transducers

8:00 a.m. – 12:00 p.m.

Douglas G. Wildes and L. Scott Smith
GE Global Research Center - Niskayuna, NY, USA

This course will provide an introduction to the design,

fabrication, and testing of medical ultrasound transducers.
Starting from an overview of the basic types of phased-ar-
ray transducers (linear, convex, sector), we will discuss how
the design for a probe is derived from its target application
and how equivalent-circuit, finite-element, and acoustic field
models can be used to optimize the design and accurately
predict performance. A discussion of the structure of an ul-
trasound probe will lead to a survey of the different types of
materials used in probes and their critical properties. Typi-
cal fabrication processes will be introduced and common
problems in probe manufacturing will be summarized. Meth-
ods for evaluating completed transducers will be discussed.
The course will highlight recent developments in probe tech-
nology, including. single crystal piezoelectrics, cMUT trans-
ducers, multi-row and 2D arrays, and electronics in probes,
and will discuss performance advantages and fabrication
difficulties which may be associated with each.
Douglas G. Wildes is a physicist with GE Global Re-
search. He earned an A.B. in physics and mathematics from
Dartmouth College and a Ph.D. in low-temperature physics
from Cornell University, then joined GE in 1985. Since 1991,
Dr. Wildes’ research has focused on aperture design, fabri-
cation processes, and high-density interconnect technology
for multi-row and 2D transducers for medical ultrasound. Dr.
Wildes has 19 issued patents and 18 external publications.
He is a member of the American Physical Society and a
Senior Member of the IEEE.
15
 L. Scott Smith is a physicist with GE Global Research.
He earned B.S. and Ph.D. degrees in physics from the Uni-
versity of Rochester and the University of Pennsylvania re-
spectively. Joining GE in 1976, he developed phased array
probes for medical ultrasound. More recently, he examined
novel probe materials and led projects on pediatric
endoscopes and adaptive acoustics. Dr. Smith has 37 is-
sued patents and over 35 refereed publications. He is a
member of the American Physical Society and a Senior
Member of the IEEE where he serves as Vice Chair for Trans-
ducers on the Ultrasonics Symposium’s Technical Program
Committee.

2 - Elasticity Imaging: Principles, Systems,

Approaches and Applications

8:00 a.m. – 12:00 p.m.

Stanislav Emelianov
University of Texas - Austin, Texas, USA

Elasticity imaging is rapidly evolving into a new diag-

nostic and treatment-aid tool. The primary purpose of this
course is to provide both a broad overview and comprehen-
sive understanding of elasticity imaging, and, as such, it is
well suited for both newcomers and active researchers in
the field. Starting with a historical introduction to elasticity
imaging, we begin to lay a foundation for static and dynamic
approaches in elasticity imaging with a brief discussion of
theory of elasticity including both the equation of equilibrium
and the wave equation. We will also review the mechanical
properties of soft tissues. Then, experimental aspects of elas-
ticity imaging will be discussed with emphasis on data cap-
ture, signal and image processing algorithms to measure
internal tissue motion induced by either internally or exter-
nally applied forces. Motion tracking methods will be intro-
duced, and techniques to increase and optimize signal-to-
noise ratio in strain imaging will be overviewed. Finally, tech-

16
niques to map elasticity and other mechanical properties of
tissue will be presented and discussed. Following an over-
view of elasticity imaging, the ultrasound elasticity imaging
techniques and their biomedical and clinical applications will
be presented. Advantages and limitations of each approach
will be discussed and contrasted with other elasticity imag-
ing techniques such as MRI or optical elastography. The
course will conclude with overview of several experimental
and commercial systems capable of ultrasound elasticity
imaging, and discussion of current and potential clinical ap-
plications of elasticity imaging.
Stanislav Emelianov received the B.S. and M.S. de-
grees in physics and acoustics in 1986 and 1989, respec-
tively, from the Moscow State University, and the Ph.D. de-
gree in physics in 1993 from Moscow State University, and
the Institute of Mathematical Problems of Biology of the
Russian Academy of Sciences, Russia. In 1989, he joined
the Institute of Mathematical Problems of Biology, where he
was engaged in both mathematical modeling of soft tissue
biomechanics and experimental studies of noninvasive vi-
sualization of tissue mechanical properties. Following his
graduate work, he moved to the University of Michigan, Ann
Arbor, as a post-Doctoral Fellow in the Bioengineering Pro-
gram, and Electrical Engineering and Computer Science
Department. From 1996 to 2002, Dr. Emelianov was a Re-
search Scientist at the Biomedical Ultrasonics Laboratory at
the University of Michigan. During his tenure at Michigan,
Dr. Emelianov was involved primarily in the theoretical and
practical aspects of elasticity imaging. Dr. Emelianov is cur-
rently an Assistant Professor of Biomedical Engineering at
the University of Texas, Austin. His research interests are in
the areas of medical imaging for therapeutics and diagnos-
tic applications, ultrasound microscopy, elasticity imaging,
photoacoustical imaging, cellular/molecular imaging, and
functional imaging.

17
3 - Ultrasound Contrast Agents: Theory and
Experimental Results

8:00 a.m. – 12:00 p.m.

Nico de Jong and Michel Versluis
Erasmus MC - Rotterdam, the Neth. & University of
Twente - Enschede, the Netherlands.

The course consists of 6 main topics: a) First an over-

view will be presented of the (clinical and pre-clinical avail-
able) contrast agents, including the properties and charac-
teristics of the gas inside the bubble and the shell surround-
ing it. b) Models of the behavior of small bubbles in a ultra-
sound field will be discussed. Simple models based on a
one dimensional mass-spring system and more complicated
models including gas and shell properties. c) Experimental
ultrasound methods for UCA will be presented for charac-
terizing the bubbles in a UCA, like harmonic and subharmonic
scattering, absorption and attenuation. Also the influence of
ambient pressure, temperature and gas concentration will
be discussed. d) Experimental optical methods for charac-
terizing individual bubbles. e) Imaging methods for contrast
agents, like fundamental, harmonic, subharmonic and
superharmonic and multi-pulse methods like pulse inversion,
power modulation etc. and new methods like chirp excita-
tion. f) Ultrasound mediated drug delivery: Interaction be-
tween mammalian cells and ultrasound in the vicinity of
bubbles will be discussed.
Nico de Jong graduated from Delft University of Tech-
nology, The Netherlands, in 1978. He got his [Link]. in the
field of pattern recognition. Since 1980, he has been a staff
member of the Thoraxcenter of the Erasmus University
Medical Center, Rotterdam, The Netherlands. At the Dept.
of Biomedical Engineering, he developed linear and phased
array ultrasonic probes for medical diagnosis, especially
compound and transesophageal transducers. In 1986 his
interest in ultrasound applications shifted toward the theo-

18
retical and practical background of ultrasound contrast
agents. In 1993 he received his Ph.D. for “Acoustic proper-
ties of ultrasound contrast agents.” Currently he is interested
in the development of 3-D transducers and fast framing cam-
era systems. De Jong is the project leader of STW and FOM
projects on ultrasound contrast imaging and drug delivery
systems. Together with Folkert ten Cate, MD, he is orga-
nizer of the annual European Symposium on Ultrasound
Contrast Imaging, held in Rotterdam and attended by ap-
proximately 175 scientists from all over the world. Since 2003
Nico de Jong is part-time professor at the University of
Twente.
Michel Versluis graduated in Physics in 1988 at the
University of Nijmegen, the Netherlands, with a special in-
terest in Molecular Physics and Astrophysics. Later, he spe-
cialized in the application of intense tunable UV lasers for
flame diagnostics resulting in a successful defense of his
PhD thesis in 1992. Michel Versluis is now a lecturer at the
University of Twente, the Netherlands, in the Physics of Flu-
ids group working on the experimental study of bubbles and
jets in multiphase flows and granular flows. He also works
on the use of microbubbles as a tool for medical diagnosis
and therapy. Dr. Versluis teaches various courses in Fluid
Mechanics, one of them focusing on the physics of bubbles.

4 - Recent Trends in Beamformation in Medical

Ultrasound

1:00 p.m. – 5:00 p.m.

Kai Thomenius
General Electric’s Corporate R&D - Niskayuna, NY,
USA

The goal of this introductory course is to review the de-

sign of ultrasound front ends and beamformers from a lin-
ear systems point of view including transduction,
beamformation, and image formation functions. We will dis-

19
cuss analytical methods used in developing the design of a
typical beamformer in use in diagnostic ultrasound today.
The key points to be covered deal with methods of analysis
of arrays and beamformers, the interaction of transmit and
receive beams with clinically relevant targets, and how this
interaction is used in image formation. The means by which
these analytical methods contribute to a beamformer de-
sign and the trade-offs involved are reviewed. The techniques
developed for such analysis will be applied to topics of cur-
rent interest involving beamformation such as system min-
iaturization, 2D arrays, synthetic aperture techniques, and
aberration correction.
Kai E. Thomenius is a Chief Technologist in the Imag-
ing Technologies Organization at General Electric’s Global
Research facility in Niskayuna, NY. His focus is on Ultra-
sound and Biomedical Engineering. Previously, he has held
senior R&D roles at ATL Ultrasound, Inc., Interspec Inc.,
Elscint, Inc., Inc as well as several other ultrasound compa-
nies, and is currently an Adjunct Professor in the Electrical,
Computer, and Systems Engineering Department at
Rensselaer Polytechnic Institute where he teaches a course
in general imaging. Dr. Thomenius’ academic background
is in electrical engineering with a minor in physiology; all of
his degrees are from Rutgers University. His long-term in-
terests have been in ultrasound beamformation and minia-
turization of ultrasound scanners, propagation of acoustic
waves in inhomogeneous media such as tissue, the poten-
tial of bioeffects due to those acoustic beams, and determi-
nation of additional diagnostic information from the echoes
that arise from such beams. Recently he has contributed to
work on coherent beamformers in millimeter wave radar
applications. He is a Fellow of the American Institute of Ul-
trasound in Medicine.

20
5 - Micromachined Ultrasonic Sensors and Actuators

1:00 p.m. – 5:00 p.m.

Ville Kaajakari, Amit Lal, and Richard White
Cornell University - Ithaca, NY and University of
California - Berkeley, CA

Part A: The goal of this part is to introduce the funda-

mentals of micromachining, and the way they affect the de-
sign and performance of ultrasonic sensors and actuators.
We will cover established micromachining techniques, such
as bulk micromachining and surface micromachining on sili-
con. Material on thin film deposition and foundries will be
presented. The relevant acoustic and ultrasonic properties
of materials used in MEMS will be discussed for predictable
device design. Nonlinearities, material property gradients,
and internal stresses will be covered to describe their effect
on design.
Part B: Case studies of sonic MEMS will be presented.
These include (1) electrostatic actuation of micromachined
membranes, nonlinearities and effective electromechanical
coupling, (2) comparison of PZT and thin-film piezoelectric
actuation of silicon bulk and surface micromachined struc-
tures (silicon horn design, microphones, speakers, flexural
plate waves, FBARS), and (4) nonlinear ultrasound in
microfluidic devices, and (5) Micro resonators for RF com-
munications.
Amit Lal is an associate professor of electrical and com-
puter engineering at Cornell University. He received his Ph.
D. in electrical engineering from the University of California,
Berkeley in 1996, and the B.S. degree from the California
Institute of Technology in 1990. Amit Lal is the leader of the
SonicMEMS group at Cornell University, which focuses on
ultrasonics, micromachining, modeling of piezoelectric sys-
tems, use of radioactive energy sources in microsystems,
and design and analysis of integrated circuits. He has pub-
lished papers on ultrasonic sensors and actuators at confer-

21
ences in ultrasonics and micromachining. He serves on the
Technical Committee on Physical Acoustics in the IEEE Ul-
trasonics, Ferroelectrics, and Frequency Control Society. He
holds patents on micromachined acoustic sources/receiv-
ers, and silicon-based high-intensity ultrasonic actuators. He
is also the recipient of the NSF CAREER award for research
on applications of ultrasonic pulses to MEMS.
Richard M. White is a professor of EECS and a found-
ing co-director of the Berkeley Sensor & Actuator Center at
the University of California at Berkeley, Dick White has con-
centrated on ultrasonics and microsensors. He has published
on thermoelastic wave generation, SAW transduction, and
flexural plate-wave sensors. He has co-authored three books
- a text for freshmen, a book on solar cells, and the refer-
ence book “Acoustic Wave Sensors”. White is a member of
the National Academy of Engineering, and has received
awards for his contributions to ultrasonics from the IEEE
and the Ultrasonics and Frequency Control societies of the
UFFC. His present research interests include ultrasonic air-
borne particulate monitoring and wireless passive proximity
metering of AC power use in dwellings.
Ville Kaajakari received his M.S. and Ph.D. degrees
in electrical and computer engineering fromUniversity of
Wisconsin-Madison in 2001 and 2002, respectively. He is
currently Senior Research Scientist at VTT Information
Technology, Finland, where his research interest is RF-
MEMS.

6 - Clinical Applications of Diagnostic Ultrasound

1:00 p.m. – 5:00 p.m.

Folkert ten Cate and Juiry W. Wladimiroff
Erasmus MC - Rotterdam, the Netherlands

The goal of this introductory course is to review the clini-

cal applications of ultrasound imaging in cardiology. The pre-

22
sentation will be illustrated with realtime 2D and 3D
Echoimages.
Folkert J ten Cate, M.D., is director of the clinical
echolaboratory of the Thoraxcenter , Erasmus MC in
[Link] main interest is in cardiomyopathies and ul-
trasound contrast both for diagnosis and treatment. He is a
Fellow of the American College of Cardiology and the Euro-
pean Society of Cardiology.
Professor Juiry W. Wladimiroff was born in The Hague.
He graduated from the school of medicine in Leiden in 1965
and was Board certified in Obstetrics and Gynecology in
1972. After some initial endeavours, Wladimiroff soon went
down to London to study with Professor Stuart Campbell at
the Post-graduate Institute at the Queen Charlotte’s Hospi-
tal. In the late 1970s he field tested Organon Teknika’s origi-
nal real-time equipments from the Netherlands (in collabo-
ration with Nicolaas Bom, the original inventor) and demon-
strated the usefulness of the MiniVisor in the rapid mea-
surement of the biparietal diameter at the bedside. His re-
search at Queen Charlotte’s started with the measurement
of fetal urinary production rate, which he continued to ex-
pand after returning to the Netherlands, looking at fetal uri-
nary production under a variety of physiological and patho-
logical situations. From then on Professor Wladimiroff be-
came particularly interested in the physiology and patho-
physiology of pregnancy and the fetus and his group was
very productive in researches pertaining to fetal cardiology,
fetal vascular and cerebral function and fetal blood flow as
assessed by doppler velocimetry. In 1974, he received his
PhD at the University of Nijmegen with a thesis on fetal
monitoring. In 1973, he started work as a consultant at the
department of obstetrics and gynecology of Erasmus Uni-
versity Rotterman at Dijkzigt Hospital; in 1977, he was ap-
pointed reader at this department, and in 1980 full profes-
sor. Since 1984, he was head of the division of prenatal di-
agnosis and since 1996, when the two divisions were
merged, head of the division of obstetrics and prenatal diag-

23
nosis at Rotterdam University Hospital. In 1981, his group
reported fetal left ventricular volume determination from a
study of two-dimensional measurement of real-time ultra-
sonic images of the left ventricle. Their group was the first to
describe doppler studies of the middle cerebral arteries and
the carotid carteries, and popularizing the carotid artery/
umbilical artery PI ratio for the assessment of fetal compro-
mise. Professor Wladimiroff was the President of the Dutch
Society of Obstetrics and Gynecology from 1993 to 1995
and was the Chairman of the National Liason Committee for
Medical Research Committees in the Netherlands. He has
organized numerous National and International Scientific
meeetings and Symposia and was the Chairman of the Edu-
cation Committee of the International Society for Ultrasound
in Obstetrics and Gynecology (ISUOG). He is also a board
member of the Society of the “Fetus as a Patient” and ex-
ecutive board member of the European Board and College
of Obstetrics and Gynecology and has carried out visitations
in departments from Slovenia to Portugal. Professor
Wladimiroff has produced over 300 important scientific pa-
pers and contributed to over 20 books and monographs. He
is well regarded by his colleagues as a great teacher and
investigator. Since 1977, he has supervised 25 PhDs on
many different aspects of prenatal diagnosis, of obstetrical,
gynecological and Doppler ultrasound, and of fetal monitor-
ing. His PhD students came from Holland, Switzerland, Brit-
ain, Indonesia and Austria. In recognition of his contribu-
tions to the advancement of ultrasound in Obstetrics and
Gynecology, he was presented the Ian Donald Gold Medal
by the ISUOG in 1997. In 1999 he received the Gold Medal
from the Drs. Haackert Foundation for “Lifetime Achieve-
ments in the field of Prenatal Diagnosis and Therapy”.

7 - Non-linear Acoustics and Harmonic Imaging

6:00 p.m. – 10:00 p.m.

Victor Humphrey
Southampton University - Southampton, UK
24
 This course will provide an introduction to the origins of
non-linear propagation, and its consequences and applica-
tions in medical ultrasound.
The first section will review the basic physics of non-
linear propagation, and discuss the propagation of plane
waves as a means of introducing non-linear acoustics ter-
minology. This will be followed by a discussion of the tech-
niques used to numerically model non-linear propagation
and the specific problems of performing measurements in
high amplitude fields with their associated distortion and
harmonic content.
The effects of diffraction and attenuation on non-linear
propagation will then be introduced by considering the fields
of transducers and arrays, and the fields they generate in
tissue; this will be illustrated by a combination of experimen-
tal results and model predictions. This will lead on to a dis-
cussion of the consequences for medical ultrasound of non-
linear propagation. Finally the application to harmonic imag-
ing will be described.
Victor Humphrey is a Professor of Acoustics at the
Institute of Sound and Vibration Research (ISVR) in
Southampton, U.K. He received his BSc and PhD degrees
from the University of Bristol in 1975 and 1981 respectively.
He then moved to the School of Physics at the University of
Bath where was promoted to Senior Lecturer. In 2004 he
took up his current position at ISVR. His initial research was
in the area of laboratory applications of non-linear paramet-
ric arrays in underwater acoustics. For this work he was
awarded the Institute of Acoustics A.B. Wood Medal 1988.
Subsequently he helped to develop a research programme
on the non-linear propagation of ultrasound in medical fields
that investigated these fields both numerically and experi-
mentally. He was awarded the University of Bath Mary Tasker
Award for excellence in teaching in 1995.

25
8 - Finite Element Modelling of Ultrasound
Applications

6:00 p.m. – 10:00 p.m.

Paul Reynolds and John Mould
Weidlinger Associates Inc - Los Altos, USA

Finite Element Modelling (FEM) of ultrasound devices

has been growing since its early use in the late 1980s and
early 1990s. It is now common, and likely soon to be univer-
sal, to find industrial and academic groups with a significant
ultrasound research component to utilize computer simula-
tions of one form or another in their work. While the research-
ers are experts in their own field of ultrasound, they rarely
have such extensive knowledge of the field of finite element
modelling and consequently often have difficulty in making
appropriate choices and decisions regarding their model-
ling needs and approach. It is the aim of this course to edu-
cate the ultrasound expert in the important considerations
with regards to the finite element modelling of ultrasound
applications, with particular emphasis on phenomena oc-
curring in front of the transducer. By the end of the course, it
is our intention that the attendees will have the basic infor-
mation on finite element simulations, and several common
but varied applications, with which to make informed deci-
sions in regards to simulating their own particular problems,
and therefore make best use of the resources available to
them. The course will be divided in four parts.

Part 1: Finite Element Basics

The first section will involve an introduction to the field
of finite element modelling, in order to ensure that all partici-
pants are aware of the basic assumptions inherent in the
various modelling approaches. Common terminology, types
of analysis (harmonic, transient etc), types of solution meth-
ods (implicit, explicit), types of numerical solver (direct, it-
erative), and typical boundary conditions, such as symme-

26
try and infinite element (e.g. absorbers and PML), validation
and verification methods will be detailed.

Part 2: Wave Propagation

The second section will concentrate on the modelling
of wave propagation through various media. Initial consider-
ation will be given to the simple, linear, elastic cases and
then moving to include the effects of long distance propaga-
tion, material discontinuities, frequency dependant attenua-
tion, and non-linearity (such as is prominent in higher-har-
monic imaging).

Part 3: Ultrasound Applications

The third section will consider a variety of ultrasound
applications. This includes accidental or intentional tissue
heating (such as with HIFU). Appropriate and accurate cal-
culation of thermal generation (sometimes called the Bioheat
Equation) and its application as a load to a thermal model
will be detailed. Aspects of ultrasound/thermal coupling will
be compared to acoustic radation force calculation, which
bear significant resemblance in approach.

Part 4: Efficient Application of Modelling Software on

Available Hardware
The ability to economically answer the questions posed
often marks the difference between a successful and a failed
project. We cover simple and effective approaches for en-
suring maximum return on time invested in FEM, and impor-
tant considerations for ensuring sufficiently accurate an-
swers. This will then extend to discussion of numerical opti-
mization techniques, and the relative costs compared to the
potential benefits considered. We will discuss common com-
puter architectures, the 32 to 64 bit transition, and multi-
processing, in order to leave the potential user somewhat
more comfortable in this rapidly changing and bewildering
field.

27
 Paul Reynolds received [Link] in Electrical and Me-
chanical Engineering from the University of Strathclyde,
Scotland, in 1994, and Ph.D in 1998 for his work on finite
element modelling of piezoelectric transducers. Since 1999
he has worked at WAI using the PZFlex finite element pack-
age to model a wide range of ultrasound and piezoelectric
applications, including medical imaging, therapeutics, SO-
NAR, and sensors.
John Mould received [Link]. and [Link]. in Civil Engi-
neering from Virginia Tech in 1978 and 1979 respectively.
He received a Ph.D. in Civil Engineering from the Univer-
sity of Colorado in 1983. Since joining WAI in 1983 he has
been an analyst and a major contributor to the develop-
ment of the entire FLEX family of codes for Nonlinear
Solids, Acoustics, Thermal, Piezoelectric and Electromag-
netic analyses.

9 - Flow Measurements and Doppler

6:00 p.m. – 10:00 p.m.

Hans Torp
Norwegian University of Science and Technology -
Trondheim, Norway

This course provides basic understanding of physical

principles and signal processing methods for flow measure-
ments and visualization; with emphasis on Doppler meth-
ods and blood flow applications. The course starts with an
overview of currently used techniques for velocity estima-
tion in pulsed and continuous wave Doppler and color flow
imaging. Statistical models for the received signal, as well
as commonly used velocity and flow estimators are devel-
oped. Several different simulation methods for ultrasound
signals from moving blood and clutter signals will be dis-
cussed. This includes fast simulation methods, as well as
full 3D point scatter models using spatial impulse response
techniques or k-space analysis. Efficient simulation tools to

28
explore estimator properties are derived, and examples on
implementation in Matlab will be shown. Methods to sup-
press clutter signals from slowly moving targets, including
regression filter will be discussed. Elements from classical
estimation theory will be applied to develop minimum vari-
ance velocity estimators in the presence of clutter noise.
The performance will be compared with commonly used
approaches for clutter rejection and velocity estimation, and
practical implementations will be discussed. Velocity com-
ponents transversal to the ultrasound beam can not be mea-
sured by Doppler techniques. However, several approaches
to overcome this limitation has been proposed, including
speckle tracking, transit time measurements, and lateral
beam modulation. Principles and practical limitations will be
discussed. Methods for visualisation of 2D vector flow infor-
mation will be shown.
Hans Torp received the MS degree in mathematics in
1978, and the Dr. Techn. degree in electrical engineering in
1992; both from the University of Trondheim, Norway. Since
1980 he has been working with ultrasound technology ap-
plied to blood flow measurements and imaging at the uni-
versity of Trondheim, in cooperation with GE-Vingmed Ul-
trasound. He is currently professor of medical technology at
the Norwegian University of Science and Technology, and
has since 1987 given courses on ultrasound imaging and
blood flow measurements for students in electrical engineer-
ing and biophysics. His research interests includes statisti-
cal signal- and image processing with applications in medi-
cal ultrasound imaging.

29
 PLENARY SESSION

Monday, September 19
Willem Burger Zaal
8:30 a.m.-10:00 a.m.

Welcome
Ton van der Steen, General Chair
John Hossack, Technical Program Chair

Awards and Recognitions

Achievement Award
Distinguished Service Award
Rayleigh Award
Outstanding Paper Award
Fellow Awards
Distinguished Lecturer Award
David Sarnoff Award

Nanoscience, From Single-Molecule Science to

Applications Cees Dekker
Delft University of Technology, Kavli Institute of
NanoScience,
Lorentzweg 1, 2628 CJ Delft, The Netherlands
[Link]

[Link]. Cees Dekker (1959) is a professor in Molecu-

lar Biophysics at the Technical University Delft. His research
takes place at the boundaries of nanotechnology, biology
and physics. His work in carbon nanotubes is most famous,
but his current interests have diverted to nanotechnology of
living cells. He published more than 130 peer reviewed pa-
pers, of which more than 20 in Nature en Science. In 2003
he was appointed as a member of the Royal Dutch Acad-
emy of Arts and Sciences. Dekker received several honours

30
and awards, including the 2001 Agilent Europhysics Prize,
the 2003 Spinoza Award of the Netherlands Organization
for Scientific Reasearch and an honorate doctorate in
Diepenbeek, Belgium. Further information can be found on:
[Link]/user/dekker/[Link]
This talk will give a broad introduction and overview of
nanoscience and nanotechnology. I will illustrate the excite-
ment and potential of this emerging field by focusing on a
number of examples, in particular carbon nanotubes for
nanoelectronics and single-molecule techniques for biophys-
ics of DNA-enzyme interactions.

EXHIBITS

Time: Daily from 8 a.m. to 6 p.m.

Location: Foyer room 2, Jurriaanse Zaal

(first floor)
Foyer room 3, Van der Mandele Zaal
(first floor)

Acqiris SA-Data Conversion Instruments

Advanced Modular Sputtering, Inc. (AMS)
Ferroperm Piezoceramics A/S
HITACHI Medical Systems
HONDA Electronics Co. Ltd.
IEEE-UFFC Society
Imasonic
Lecoeur Electronique
Microfine Materials Technologies PTE LTD
Precision Acoustics Ltd.
Smart Material GmbH
Sonora Medical Systems, Inc.
Sound Technology, Inc.
Valpey Fisher Corporation
W. L. Gore & Associates

31
 MONDAY, SEPTEMBER 19, 2005

*Author presenting paper

POSTER SESSIONS

Session: PS
STUDENT PAPER FINALISTS
Chair: Helen Routh, Philips Medical Systems,
and Tom Thomas, Siemens Medical Solutions

PS-1
HIGH FREQUENCY ULTRASOUND
CHARACTERIZATION OF THE BLOOD CLOTTING
PROCESS: INTRA- AND INTER-INDIVIDUAL
VARIATIONS.
R. LIBGOT*1, F. OSSANT1,2, Y. GRUEL2, and F. PATAT1,2, 1LUSSI FRE 2448
CNRS, Tours, France, 2Bretonneau University Hospital, Tours, France.
Corresponding e-mail: libgot_r@[Link]
In a previous study [1], we presented a device capable to simultaneously estimate
high frequency ultrasound parameters both in double-transmission mode (DT-
mode) and in backscattering mode (BSC-mode). Sound velocity was estimated
in DT-mode and BSC coefficient (from which the integrated BSC coefficient
(IBC) and the effective scatterer size (ESS) were estimated) in BSC-mode. We
used a 45 MHz single element transducer for the DT-mode, and a real time high
frequency echographic devices allowing spectral analysis of signals in the
frequency range 10-30 MHz (Dermcup 2020TM). We showed that BSC-mode
and DT-mode parameters were complementary to give a fine description of
blood coagulation process in terms of mechanical properties (blood solidification)
and spatial distribution of blood components (formation RBC aggregates).
In the present paper, our purpose was to estimate interindividual and
intraindividual variations of these ultrasound parameters in order to assess the
clinical potential of this technique. Blood samples were collected from 10 healthy
donors to study interindividual variations. To study intraindividual variations, blood
samples were collected 3 times over 24 hours from each volunteer. Several
protocols were then conducted: samples were taken in 3.5 ml sodium citrate
tubes and placed in a temperature controlled chamber (37°C ± 0.1) for 15 and
30 minutes before the start of the experiment (tcit), the coagulation process was
initiated by adding 100 and 200 µl of CaCl2 (with concentration of 0.5M). The
hematocrit and fibrin concentration were determined for each blood sample.
During the blood clotting process (two hours’ observation), the entire set of
acoustic parameters was measured with 15 seconds time resolution for the DT-
mode and 30 seconds for the BSC-mode.

32
In the DT-mode, the results were focused on the variation kinetics of the acoustic
velocity. Three characteristics could be considered to differentiate these kinetics:
the time t0 at which the velocity increased strongly (typically between 20 to 50
minutes), the slope S of this increase (typically between 0.013 and 0.029 m/
s^2) and the final increase I of the velocity (typically between 20 and 35 m/s). In
the BSC-mode, the IBC varied as the fourth power of the frequency (Rayleigh
scattering) during the first few minutes of the process and then decreased, from
4 to 2.5, as the clot formed (corresponding to ESS values between 0 and 25 µm
for a Gaussian model).
Preliminary results from RF-signals in the frequency range 30-60 MHz with a 50
MHz center frequency probe were also obtained and compared to those obtained
for 10-30 MHz. Histological analysis of the clots were performed and several
acoustic parameters values were analyzed in relation to clot structures.
Finally, the kinetics of high frequency ultrasound parameters, for instance the
acoustic velocity, during blood clotting process is a promise technique to
characterize clotting properties of whole blood.
1:Ossant and al., High frequency ultrasound characterization of the coagulation
process of whole blood, Ultrasonics Symposium, 2004 IEEE pp846-849

PS-2
ULTRASONIC DETECTION OF THE ANISOTROPY OF
PROTEIN CROSS LINKING IN MYOCARDIUM.
S. BALDWIN*, M. YANG, K. MARUTYAN, K. WALLACE, M. HOLLAND, and J.
MILLER, Washington University in St. Louis, St. Louis, MO.
Corresponding e-mail: slb@[Link]
BACKGROUND: Diastolic dysfunction may arise, in part, because of an increase
in myocardial stiffness from cross-linking of extracellular matrix proteins such
as collagen. Studies by others indicate that high-frequency (30 MHz to 50 MHz)
ultrasound can detect chemically-induced increases in protein cross-linking (C.
S. Hall et al., “High-Frequency Ultrasound Detection of the Temporal Evolution
of Protein Cross-Linking in Myocardial Tissue,” IEEE Trans. Ultrason., Ferroelect.,
Freq. Contr., vol. UFFC-47, pp.1051-1058, 2000).
OBJECTIVE: The goal of the current study was to measure, as a function of the
angle of insonification relative to that of the predominant myofiber orientation,
changes in myocardial attenuation resulting from increased protein cross-linking.
METHODS: Through-transmission radiofrequency-based measurements of the
attenuation coefficient (3 MHz to 7 MHz), for angles of insonification over a
complete rotation relative to the predominant myofiber orientation in 5 degree
increments, were performed at room temperature on 36 specimens from 12
freshly excised lamb hearts. These specimens were then fixed in formalin to
induce protein cross-linking. After sufficient time had elapsed to ensure that the
process of cross-linking had reached completion, measurements were repeated
on the identical specimens.
RESULTS: For both freshly excised and formalin-fixed myocardium, the
attenuation coefficient was found to increase as a function of frequency in an

33
approximately linear manner. The attenuation varied systematically as a function
of angle of insonification, with a minimum perpendicular to, and a maximum
parallel to, the direction of the myofibers. Increased protein cross-linking inferred
from fixation resulted in a maximum increase in attenuation at perpendicular
insonification. For angles near perpendicular to the predominant direction of
the myofibers, the measured slope of attenuation increased from 0.52 ± 0.07
dB/(cm⋅MHz) (mean ± one standard deviation) for freshly excised to 0.85 ± 0.08
dB/(cm⋅MHz) for formalin-fixed myocardium. In contrast, for parallel insonification
uncertainties exhibit considerable overlap (1.88 ± 0.17 for freshly excised and
1.75 ± 0.19 dB/(cm⋅MHz) for formalin-fixed myocardium).
CONCLUSION: This study indicates that effects of fixation, and therefore
presumably protein cross-linking, in myocardium can be anisotropic, suggesting
that the response of the extracellular collagenous matrix to changes in cross-
linking is directionally dependent. The anisotropy of ultrasonic attenuation may
thus provide an approach for the noninvasive monitoring of the extent and
progression of myocardial disease associated with changes in protein cross-
linking.
NIH R37 HL40302

PS-3
DETECTION OF THE MYOCARDIAL BOUNDARY IN
THE LEFT VENTRICLE FROM SIMULTANEOUSLY
ACQUIRED TRIPLANE ULTRASOUND IMAGES USING
MULTI VIEW ACTIVE APPEARANCE MOTION MODELS.
J. HANSEGÅRD*1, S. URHEIM2, E. STEEN4, H. TORP3, B. OLSTAD3, S. MALM3,
and S. RABBEN4, 1University of Oslo, Oslo, Norway, 2Rikshospitalet University
Hospital, Oslo, Norway, 3The Norwegian University of Science and Technology,
Trondheim, Norway, 4GE Vingmed Ultrasound, Horten, Norway.
Corresponding e-mail: jogerh@[Link]
We report a new algorithm for detecting the myocardial boundary in the left
ventricle (LV) from simultaneously acquired triplane ultrasound (US) image
sequences using Multi View Active Appearance Motion Models (MVAAMM).
Coupled boundary detection in three planes can potentially increase the accuracy
of LV volume measurements, and also increase the robustness of the boundary
detection over traditional methods.
The MVAAMM is an extension of the Active Appearance Motion Models (AAMM)
and is capable of generating LV triplane image sequences. The parameters of
the model are optimized iteratively to fit the patient data set, and the myocardial
boundary is then extracted from the fitted model. Shape and texture are coupled
between the image planes, while the model pose is fitted independently.
A database of triplane image sequences from full cardiac cycles, including the
standard apical four chamber, two chamber, and long axis views were established
from 20 volunteers, including 12 healthy persons and 8 persons suffering from
heart disease. Patient cases with asynchronies were not included. A Vivid 7
scanner (GE Vingmed Ultrasound, Horten) was used for acquisition of the data
sets. For each dataset the LV myocardial boundary was manually outlined, and
34
the end diastolic (ED) and end systolic (ES) frames were determined visually
for phase normalization of the cycles.
Evaluation of the MVAAMM was done by detecting the LV myocardial boundaries
in the test sets. Variation within the data sets is critical when training the model.
Therefore, a leave one out approach was used for evaluation, resulting in 20
models where one case was left out from each model. The point distance between
the automatically detected myocardial boundary and the manually outlined
boundary was found as the mean of the distances between corresponding points
along the contours. The detected volumes at ED, mid systole and ES (VolC)
were compared to manually determined volumes (VolM). Based on the ED and
ES volumes, the Ejection Fraction (EF) was calculated.
Table 1 shows the mean and 1 standard deviation of the point distance, volume
error, volume fractional error, EF error, and also the regression line for VolC as
a function of VolM. The detected volumes at ED, mid systole and ES correlate
well with the manually determined volumes (R2 = 0.87). The correlation between
detected EFs and manually determined EFs was poor (R2 = 0.29), this is probably
caused by the method’s tendency to identify an average motion pattern.
Dropouts are handled well by the MVAAMM, because the shapes and textures
of the three planes are coupled. The views with the largest point distances have
one or more foreshortened views, resulting in a suboptimal detection. Adding
more samples to the training database or coupling of the model’s pose
parameters between the views may improve the performance in such cases.
Table 1: MVAAMM Results

Mean Point distance (mm) 4.1 ± 1.9

Error (VolM - VolC) (ml) 7.0 ± 14
Volume fractional error (1 - VolC/VolM) (%) 13 ± 12
Volume regression (y - VolM) (ml) y = 1.0 x - 11
Volume correlation coefficient (R2) 0.87

We wish to acknowledge Jan Yee at GE Vingmed Ultrasound for gathering

ultrasound data. The project is supported by The Research Council of Norway.

PS-4
3D PERFUSION MAPPING IN THE INTACT MOUSE
HEART AFTER MYOCARDIAL INFARCTION USING
MYOCARDIAL CONTRAST ECHOCARDIOGRAPHY.
Y. LI*, Z. YANG, B. A. FRENCH, and J. A. HOSSACK, University of Virginia,
Charlottesville, VA.
Corresponding e-mail: yinbo@[Link]
Due to the widespread availability of transgenics and knock-outs, the murine
species provides a very useful and robust model in which to study human
35
cardiovascular disease. Myocardial contrast echocardiography (MCE) using
microbubbles as contrast agent has been shown to be an effective tool in
characterizing myocardial perfusion in vivo, and it has been employed previously
for two-dimensional (2D) myocardial perfusion analysis in mice. However, three-
dimensional (3D) analysis, fully encompassing the entire left ventricle (LV), is
required for a more complete understanding of the relationship between LV
perfusion and function after myocardial infarction (MI). In this work, we studied
an intact mouse model of surgically-induced MI resulting from a Left Anterior
Descending (LAD) occlusion. Healthy mice which underwent a similar invasive
procedure but with no occlusion were also studied as controls. Imaging was
performed at 14 MHz using a non-destructive, contrast-specific, imaging mode
(‘Contrast Pulse Sequences’ [1]). The infusion of contrast agent was
accomplished at a slow and steady flow rate using a syringe pump driving a 1ml
syringe. For each mouse heart, MCE images were acquired in parallel short
axis cross-sections of the heart at 1 mm elevational increments. For accurate
3D reconstructions, calibrated ECG gating and a tri-axial adjustable micro-
manipulator were used for temporal and spatial registration, respectively.
Perfusion analysis was conducted by analyzing the local refilling processes
within the LV myocardium. This analysis indicated a significantly slower flow
rate in areas of perfusion defects compared to normal areas. MCE images
acquired during steady-state perfusion of each 1mm slice of the LV were color-
coded to indicate relative perfusion, with blue corresponding to well-perfused
areas and red corresponding to perfusion defects. As a standard for comparison,
post-mortem tissue was stained with Phthalo blue and TTC red dyes to identify
perfused and ischemic yet viable LV myocardium, respectively. A good correlation
(R 2 > 0.87) was observed between the MCE-based, in vivo measurements and
the tissue-based, ex vivo measurements of % perfused area in each tissue
slice. 3D multi-slice models and 3D volumetric models of the distribution of
perfusion in each LV were created for improved visualization and quantification
of the 3D extent of the perfusion defects. These models demonstrated a promising
match with the results of post-mortem tissue staining.
[1] P. Phillips, “Contrast Pulse Sequences (CPS): Imaging non-linear
microbubbles,” Proceedings of the 2001 IEEE Ultrasonics Symposium, vol. 2,
pp. 1739-1745, 2001.
NIH grants EB001826. EB002349, Siemens Medical Solutions, Mountain View,
CA and Vermon SA, France.

PS-5
APPLYING IN VITRO ELASTICITY IMAGING RESULTS
TO OPTIMIZE IN VIVO BREAST LESION
CHARACTERIZATION USING A COMBINED 3D US/
DIGITAL X-RAY SYSTEM.
R. BOOI*1, P. CARON1, R. ERKAMP1, H. XIE1, A. KAPUR2, G. LECARPENTIER1,
M. ROUBIDOUX1, J. FOWLKES1, and M. O’DONNELL1, 1University of Michigan,
Ann Arbor, 2General Electric Global Research Center, Schenectady, NY.
Corresponding e-mail: rbooi@[Link]

36
Ultrasound-based reconstructive elasticity imaging has great potential for
diagnosis and characterization of breast lesions. Applying external strain with a
mammographic paddle as part of a combined 3D US/Digital X-ray system
provides more uniform deformation and breast stability, offering opportunities to
improve image fidelity. In this study, we examined in vitro and in vivo strain
image quality with three GE transducers (M12L, 10L, 7L) each operating at
several frequencies between 5-10 MHz and four TPX paddle thicknesses to
predict optimal in vivo results with the combined system. Out-of-plane motion
was measured by translating an ultrasonic transducer across a breast phantom
(ATS, model BB-1) in 50µm steps over 400µm. Each image was correlated to
the first in the sequence using conventional 2D phase-sensitive speckle tracking
techniques to determine the rate of elevational decorrelation. Next, in-plane,
strain-limited decorrelation was evaluated by compressing the phantom at 0.15%
increments up to 5% strain. Images were correlated at step sizes 0-7 (0-1.05%
mean strain) using two-pass 2D speckle tracking algorithms and estimated
displacements were accumulated. Adaptive strain estimation (ASE) was applied
to maximize CNR throughout the final strain image. Overall, the 10L transducer
caused the least decorrelation due to out-of-plane motion (R = 0.97 at 7.5MHz
and 400µm elevational translation). In-plane decorrelation was also primarily
strain-limited with the 10L transducer at 7.5MHz, with R = 0.9 for a 1.6% strain
step. Accumulated strain images after ASE demonstrated a CNR = 3.8 with the
M12L transducer at 10 MHz. Using the same post-processing techniques, the
10L transducer at 7.5 MHz demonstrated a CNR = 4.7. Of the four TPX paddle
thicknesses (0, 0.25, 1.0, 2.5 mm) used in the in vitro study, the 2.5mm paddle
created clearer strain images with less artifacts than the thinner paddles under
the same experimental conditions, providing more uniform deformation during
compression by bowing less than the thinner paddles. Next, we evaluated sources
of in vivo chest wall motion in 7 subjects to minimize patient motion during the
scan. Patients were asked to breathe shallowly during exams as it caused less
decorrelation due to chest wall motion (Ravg = 0.96 over 91 frames) than holding
their breath (Ravg = 0.93). In vivo results were acquired with the 10L at 7.5 MHz
and TPX 2.5mm paddle using continuous compression over 2.1 seconds. Strain
images clearly distinguished between tissue types when accumulated up to 4%
strain and ASE was applied. In vivo results are limited by out-of-plane motion
and CNR but are expected to improve significantly with full 3D tracking. These
early successes indicate that using elasticity imaging with the combined system
shows great potential for characterizing breast lesions that might be malignant,
reducing the need for biopsies.
This work was supported in part by NIH grant RO1 CA 091713.

PS-6
STRESS FIELD FORMATION FOR MULTIFREQUENCY
VIBRO-ACOUSTOGRAPHY.
M. URBAN*1, G. SILVA2, R. KINNICK1, M. FATEMI1, and J. GREENLEAF1, 1Mayo
Clinic College of Medicine, Rochester, MN, 2Universidade Federal de Alagoas,
Maceio, AL, Brazil.
Corresponding e-mail: [Link]@[Link]

37
Background: Vibro-acoustography is a recently developed method that uses
the dynamic radiation force (stress) of ultrasound to locally excite an object.
Presently, to form the stress field we typically use two ultrasound beams whose
frequencies differ by ∆f, typically in the kilohertz range. The two beams interfere
at the system focus creating the radiation stress at the difference frequency ∆f.
The system point spread function (PSF) is given as a function of the radiation
stress. The object response to the radiation stress may vary significantly as ∆f
changes due to its internal resonances. Thus, acquiring images with multiple ∆f
values in one scan would provide more information than using one difference
frequency. Methods: We propose a multifrequency vibro-acoustography method
that uses a multifrequency radiation stress produced by an array transducer
driven with N ultrasound frequencies. The N ultrasound beams interfere at the
focus of the transducer producing up to N(N-1)/2 unique difference frequencies
in the radiation stress. This method can produce a multifrequency image at the
different N(N-1)/2 frequencies with only one scan of the region of interest,
increasing the information of the scanned region by a factor of N(N-1)/2. The
objective of this paper is to present the image formation theory in multifrequency
vibro-acoustography systems with experimental validation. The radiation stress
generated by sector array and annular array transducer systems is analyzed
theoretically. For experimental validation, an eight element sector array
transducer driven with four different ultrasound frequencies is used to produce
a PSF with six components at different ∆f values. A stainless steel sphere with
diameter 0.51 mm embedded in a gelatin phantom was scanned to obtain a
measurement of the PSF. A Doppler laser vibrometer is used to measure the
vibration velocity of the sphere, which is proportional to the radiation stress
upon the sphere. The detected signal was processed using a bandpass filter
and lock-in amplifier to extract each of six PSF components. In another
configuration, the raw signal from the laser was digitized for off-line digital filtering
to insure that each PSF component can be extracted from the raw data set. We
compared the measured and the theoretical PSF components through their
spatial distribution, mainlobe width at -6 dB, and sidelobe levels. Lastly, we
used a computer phantom with small spheres to demonstrate the usefulness of
multifrequency vibro-acoustography for microcalcification detection in breast
imaging. Results: The main lobe width numerically evaluated for a PSF
component is 0.91 x 2.06 mm, while the experimental measurement yields 0.98
x 1.94 mm. The theoretical and measured sidelobe levels are -7.29 dB and -
7.67 dB, respectively. Conclusion: The proposed method holds the potential
for a large gain of information with no increase in same scanning time when
applied to conventional vibro-acoustography systems.
This study was supported in part by grants EB002640, EB002167, and EB00535-
03 from National Institutes of Health, and DCR2003.013 from FAPEAL/CNPq,
Brazil.

38
PS-7
A LATERAL FIELD EXCITED ACOUSTIC WAVE
PESTICIDE SENSOR.
W. PINKHAM*, L. FRENCH, D. FRANKEL, and J. VETELINO, University of
Maine, Orono, ME.
Corresponding e-mail: [Link]@[Link]
Organophosphate pesticides are commonly used worldwide to control pests in
fruits and vegetables. Although these pesticides serve their purpose in protecting
crops, excessive pesticide use can have adverse effects on the environment
and also jeopardize the health of the consumer. Government agencies in the
U.S. limit the amount of pesticides that may be used, however many agricultural
products are imported from countries where pesticide use may not be regulated.
Therefore, to minimize environmental effects and protect the health of consumers,
a need exists for a low-cost portable sensor that will detect harmful pesticide
levels. Recently a lateral field excited (LFE) acoustic wave sensor on AT-cut
quartz has been developed [1] which is capable of detecting both mechanical
and electrical property changes in a film which selectively sorbs a target analyte.
In the LFE sensor the transverse shear mode (TSM) is excited by electrodes on
the reference surface and the sensing surface is unmetallized and coated only
with the target analyte selective film, resulting in a much simpler structure than
the standard AT-cut quartz crystal microbalance (QCM). In the present work a
polyepichlorohydrin (PECH) film is evenly spun onto the sensing surface of an
LFE sensor to detect the organophosphate pesticide, phosmet (C11H12NO4PS2),
in an aqueous environment with a flow through system. When comparing the
LFE sensor to the standard QCM as a pesticide sensor, results show that the
LFE sensor has at least 25% greater sensitivity with a lower limit in the parts per
billion (ppb) range, 12% shorter response times and more consistent
reproducibility. The improvement in sensitivity, response time and reproducibility
is most likely due to the absence of the gold film on the sensing surface. In
particular the electric field associated with the TSM is able to penetrate directly
into the analyte selective film and monitor film electrical property changes
resulting in higher sensitivity. Quicker response times and better reproducibility
may also be attributed to the absence of the gold film and any aging effects
associated with the film. In conclusion, the simplicity in structure and improved
sensing properties of the LFE sensor results in a sensor platform which may
replace the standard QCM in a wide range of applications.
[1] Y. Hu, L. A. French Jr., K. Radecsky, M. Pereira da Cunha, P. Millard, and J.
F. Vetelino. “A Lateral Fiield Excited Acoustic Wave Sensor.” IEEE Transactions
on Ultrasonics, Ferroelectrics, and Frequency Control, Vol. 15, No. 11, pp. 1373-
1380, November 2004.
*This work is supported by the National Science Foundation under grant
0330100.

39
PS-8
MORPHO-MECHANICAL ANALYSIS OF THE DENTIN-
CEMENT INTERFACE STRENGTH USING A SCANNING
ACOUSTIC MICROSCOPE.
E. BAKULIN*1, L. DENISOVA2, R. MAEV1, F. RUSANOV2, A. DENISOV2, D.
GAVRILOV2, F. SEVERIN1, and G. GRAYSON3, 1Windsor University, Windsor,
Ontario, Canada, 2 Institute for Biochemical Physics, Moscow, Russian
Federation, 3Ultradent Clinical Research Ltd., Windsor, Ontario, Canada.
Corresponding e-mail: deniso@[Link]
Up to now caries stays one of the most widely spread human diseases. A carious
tooth treatment as a rule includes restoration of the pathologically changed tissue
with a dental material cement, amalgam, resin etc. Therefore a restored tooth is
a small complicated construction formed of biological tissues and an artificial
restorative material. Mechanical strength of such a construction depends of
many factors, and one of the most significant is adhesive strength of the contact
zone between dentin and dental cement. Cement shrinkage, pores, cavities
and voids inside restoration and, particularly, in the interface between dentin
and restoration can dramatically influence upon the quality of the treatment.
The aim of the present study is to develop experimental approaches to evaluate
dentin-cement interface quality using a scanning acoustic microscope.
Human teeth extracted by orthodontic reasons in dental clinics have been used.
Special model samples imitating good and failed adhesion have been prepared.
The study has been performed with scanning acoustic microscopes Sonix
(Sonoscan, USA) and Tessonix ([Link] et al). The acoustic lens providing
ultrasound frequency 50 MHz (lateral resolution about 30 microns) has been
used. Following the investigation in an acoustic microscope, the teeth and
prepared samples have been sectioned in accordance with the scanning position
and the results, obtained with acoustic microscope have been verified using a
light microscope.
Acoustic impedance of the dental materials, which we use in the study, is quite
similar to that of dentin. Therefore in the case of a good bonding strength between
the cement and dentin almost all acoustic energy of the focused ultrasound
passes from the restoration into dentin. This is why we obtain practically no
reflected signal from this interface. If due to the shrinkage there is a void (filled
with tissue liquid or with air) in the cement-dentin interface, then the ultrasound
signal is reflecting not from dentin, but from the media with comparatively low
acoustic impedance; so the largest part of the acoustic energy will be reflected
and registered. In this case the interface between cement and dentin will became
visible in acoustic images, formed in B- or C-scan mode. A series of the acoustic
images of the restored teeth and experimental models demonstrate the difference
between the imaging of the good and failed adhesion in the cement-dentin
interface.
The morphological data obtained non-destructively in the acoustic microscopes
are in a good agreement with the results of the examining of the same samples
after sectioning in the optic microscope. The experimental approaches developed
in the present study can serve as a basis for the development of new instruments
and methods for dental clinical practice.
40
Keywords: acoustic microscopy, focused ultrasound, dental cement, cement-
dentin adhesion, nondistructive evoluation

PS-9
PARAMETRIC MODELING OF WAVE PROPAGATION
IN GAS MIXTURES - A SYSTEM
IDENTIFICATION APPROACH.
J. MARTINSSON* and J. E. CARLSON, Lulea University of Technology / EISLAB,
Lulea, Sweden.
Corresponding e-mail: [Link]@[Link]
Using ultrasonic techniques, the measurable properties of a gas are characterized
by the frequency dependent attenuation and speed of sound. The properties
are directly related to the bulk modulus by the wave number. For a non-
attenuating and non-dispersive gas the bulk modulus is a real-valued constant
containing the equilibrium density and the thermodynamic speed of sound. For
dispersive gases, the relationship between the acoustic pressure and the
condensation is dynamic, and the bulk modulus connecting these two quantities
is both complex-valued and frequency dependent.
The standard approach to describe the dynamics of the bulk modulus is to
parameterize it given the physical knowledge of the gas properties under
investigation. However, for gas mixtures with complex dynamic behavior and/or
unknown components, a complete physical description of the bulk modulus is
generally not available. For these situations a common approach is to use non-
parametric methods to describe the dynamics. Although non-parametric
techniques are easy to apply, they give only moderately accurate descriptions.
To obtain higher accuracy, parametric models must be used. When choosing
the model structure, two things are desirable: First, the model should capture
the dynamics of the system and second, the model parameters should give
information about the underlying physical properties.
In this paper, the bulk modulus is parameterized using a rational transfer function
(spring-dashpot model). This structure has a well-known connection to some
physical properties, while still keeping the number of parameters reasonably
low. We use system identification techniques to estimate the parameters and
cross-validation to prevent over-parameterization. The model is validated by
analyzing the prediction errors which show that the prediction errors are
uncorrelated with the measured echoes. This means that the parametric model
is able to capture the dynamics of the true system.
The proposed method is compared with standard non-parametric techniques
using pulse-echo measurements in ethane, oxygen, and mixtures of the two.
The experimental results show that the variances of the estimates are
considerably smaller using the proposed method compared to non-parametric
methods, especially for noisy data.
Generous grants from the Swedish Energy Agency and the Kempe Foundation
are gratefully acknowledged.

41
PS-10
INTERMEDIATE FREQUENCY RESONATORS USING
LAMB WAVES CO-INTEGRATED WITH BULK
ACOUSTIC WAVE RESONATORS.
A. VOLATIER*1,3, G. CARUYER1,3, E. DEFAY2, D. PELLISSIER TANON1, P.
ANCEY1, and B. DUBUS3, 1STMicroelectronics, Crolles, France, 2CEA-LETI/
DIHS/LCRF, Grenoble, France, 3IEMN, Lille, France.
Corresponding e-mail: volatieral@[Link]
Bulk Acoustic Wave (BAW) resonators exhibit attractive properties in term of
power handling capacity and on-chip integration to realize filters in the GHz
range. In a BAW resonator, a thin piezoelectric layer (a few µm) deposited
between two electrodes is driven in its thickness extensional mode of vibration.
To get a high quality factor, this structure is decoupled from the substrate by a
multilayer Bragg reflector (Solidly Mounted Resonator: SMR) or a surface micro-
machining membrane (Film Bulk Acoustic Resonator: FBAR). A current problem
in the design of BAW resonators is the existence of spurious resonances due to
Lamb waves propagation close to the thickness extensional mode which generate
ripple in the filter passband.
In this work, it is demonstrated that Lamb wave resonances can be used to
realize resonators in the 30-250 MHz range with high quality factor by using
specific electrode design and electrical excitation.
1D simulation (based on Mason model) is usually sufficient for the design of
BAW resonators. In our case, two dimensions behaviour of Lamb waves is
modelled using finite element ATILA code. Design of Lamb wave resonators are
performed using modal and harmonic analysis. Typical dimensions of resonator
are 100 µm width and 450 µm length (clamped-clamped beam). Influence of
electrode geometry and type of excitation (one- or two-phase) upon coupling of
fundamental or higher harmonic resonances will be emphasized.
Lamb wave resonators are realized using a surface micro-machining process
compatible with Above IC integration. Sacrificial layer is first deposited, patterned
and encapsulated by SiN film (600 nm). AlN thin film are elaborated by DC
pulsed sputtering and the active stack of the resonator is made of Ti (50 nm) / Pt
(100 nm) / AlN (550 nm) / Pt (200 nm). Lateral dimension of resonators is defined
by etching of AlN and SiN thin films. The release of the structure is made using
dioxygen plasma. According to a completely similar process Film Bulk Acoustic
Resonators (FBAR) are processed at the same time.
Symmetrical S0 Lamb wave resonances ranging from 30 MHz to 250 MHz are
measured for different electrode geometries and electrical excitations (network
analyzer HP 8753E). High quality factors (Qs and Qp) of 2000 and k2~0.8% (k
coupling coefficient) are obtained at 92 MHz for 3rd harmonic resonator. These
electrical measures agree well with 2D simulations previously described. As for
Bulk Acoustic Wave resonators, thickness extensional resonance is measured
at 2.4 GHz.
The electrical characteristics of Lamb wave resonator are potentially suitable
for channel filtering or oscillator applications and co-integrating Lamb wave and
FBAR resonators in one single Above IC process promises to offer great potential

42
for design of resonators and filters in the field of Intermediate (IF) and Radio
Frequency (RF) applications.

PS-11
CALIBRATION OF ACOUSTIC RADIATION PRESSURE
FIELD INSIDE MICROCHANNELS USING
MICROPARTICLE ZETA POTENTIAL MEASUREMENT.
M. ARAZ* and A. LAL, Cornell University, Ithaca, NY.
Corresponding e-mail: mka22@[Link]
We have previously presented a low power, small scale ultrasonic excitation of
bending waves on a glass capillary microfluidic system by a laser-cut PZT plate
[1]. Complex acoustic force field generated by the inertial coupling of standing
flexural waves on the capillary body can be used to collect and manipulate
cells, silica or polystyrene beads suspended in the fluid. As focusing of particles
from 300nm to 10 µm (or larger) size range is possible, separation of particles
with respect to size or acoustic impedance is also possible by 2-D radiation
pressure fields [1]. While the FEM, and analytical modeling of the actuator
dispersion relationships agree with the experimental results for a broad range
of frequency regime of the actuation [2], nonlinear affects due to excessive
bending leads to parametric excitation of multiple modes, making modeling of
acoustic radiation pressure difficult. Since two of the possible proposed
applications of the device are controlled biological assays and cell growth in 3-
D, a quantitative understanding of the distribution of the radiation force inside
the capillary is needed.
Related to the problem given above, we present a new method to estimate the
acoustic field inside the capillary. It is known that acoustic radiation force on a
spherical object is related to its radius. While radiation force decreases by the
size of the particle, small particles collected inside the channel tend to disperse
due to Brownian motion and electrostatic interactions of the Debye layer
surrounding them. The Debye layer interaction can be quantified by measuring
the zeta potential of the suspended particle. While Brownian motion will be only
related to the temperature and the viscosity of the suspension, varying the pH
of the solution enables control of the zeta potential around the particles. Although,
zeta potential change has negligible affect on the acoustic field, it changes the
electrostatic interaction between the particles depending on the value of the pH
change. In this sense possibility of calculating the dispersion force due to the
repelling of charged particles and Brownian motion can be used to estimate the
counter balancing ultrasonic force produced by the acoustic field. Our preliminary
results show that 500 nm silica beads suspended in pH 7 solution, having about
-30mV of measured zeta potential tend to collect in a broader range forming a
cloudy structure compared to the same particles in pH 3 solutions. The tightly
packed structures form due to smaller zeta potential. The size of the radiation-
induced force induced clusters is a sensitive function of the zeta function. Since
the radiation force experiments can be run on samples with different zeta
potentials, a quantitative value for the radiation forces can be extracted. This
extraction procedure and analysis of the limits of the procedure will be presented.

43
1) [Link], [Link], [Link], “Ultrasonic Separation in Microfluidic Capillaries”
IEEE UFFC, Honolulu, 2003
2) [Link], [Link], [Link], “Finite Element Modeling of Ultrasonic Separation
at the Microscale”,IEEE UFFC, Montreal, 2004

PS-12
CANTILEVER RESONANCE INDUCED IN SITU BY
MAGNETOSTRICTION FOR ACTIVE FLOW CONTROL.
O. DUCLOUX* 1 , N. TIERCELIN 1 , Y. DEBLOCK 1 , P. PERNOD 1 , V.
PREOBRAZHENSKY1,2, and A. MERLEN3, 1LEMAC / IEMN-DOAE – UMR CNRS
8520, Ecole Centrale de Lille, Cité Scientifique - Av Poincaré, Villeneuve d’Ascq,
France, 2LEMAC / Wave Research Center - GPI-RAS, Moscow, Russia, 3LEMAC
/ LML UMR CNRS 8107, Villeneuve d’Ascq, France.
Corresponding e-mail: [Link]@[Link]
Active flow control is at the intersection between the needs of the aerospace
industry and the possibilities of microactuators. By controlling the flow on air
wings for example, one may decrease dramatically aircraft fuel consumption
and/or noise. Such a control can be achieved for example by pulsed air blowing
through submillimetric holes distributed at specific locations of the wings of the
airplane [1]. To achieve such pulsed jets, a microvalve containing an internal
resonant cantilever actuated by a magnetostrictive film has been designed,
fabricated and characterized. The microvalve is fed by a pressurized source of
air and the resonating cantilever acts as a deflector on the internal fluid flow
addressed alternatively to the output hole on the wing’s surface or to a recycling
output.
Magnetostrictive films were chosen as actuating means for they can provide
high induced stress compared to other kinds of active materials, and also because
of their remote actuation capability. In this device, artificially nanostructured TbCo/
FeCo multilayers were used: in these layers, a large magnetoelastic coupling
can be kept while having a well-defined in-plane magnetic anisotropy, with a
relatively low saturation field. These characteristics allow the induction of a Spin
Transition Reorientation (SRT) state in the layer: the SRT corresponds to a
magnetic instability that is used to increase dramatically the magnetoelastic
sensitivity[2] and make it compatible with silicon Microsystems : that way, the
field produced by the two microcoils is sufficient for the dynamic actuation.
The actual microsystem consists in a 1000 µm long and 1.5 µm thick polysilicon
cantilever on which a nanostructured magnetostrictive film is deposited by RF
Sputtering: (TbCo/FeCo)xn, total thickness 0.5 µm to 1µm. On each edge of the
cantilever, one microcoil has been processed with 4 windings and a wire thickness
of 20 µm. Each coil produces the 20 Oe dynamic magnetic field necessary for
the actuation. The magnetic field is then transmitted from one coil to the other
via the magnetostrictive film, enabling the cantilever’s resonance at a 2 kHz
frequency.
Characterization of the resonance frequency, amplitude and magnetoelastic
coefficient are achieved by interferometric means. Outcoming flow is visualized
by conventional strioscopy method using He / Air optical indice difference. The
results of these characterizations are presented
44
[1] Kumar, S.M.; Reynolds, W.C.; Kenny, T.W.; “MEMS based transducers for
boundary layer control”, Micro Electro Mechanical Systems, 1999. MEMS ’99.
Twelfth IEEE International Conference on 17-21 Jan. 1999 Page(s):135 - 140
[2] N. Tiercelin, V. Preobrazhensky, [Link], H. Le Gall, J. Ben Youssef,
“Nonlinear actuation of cantilevers using giant magnetostrictive thin films”,
Ultrasonics 38(2000) 64-66

PS-13
DOUBLE-RESONANCE SAW FILTERS.
J. MELTAUS*1, V. P. PLESSKY2,1, and S. S. HONG3, 1Helsinki University of
Technology, Espoo, Finland, 2GVR Trade SA, Bevaix, Switzerland, 3Samsung
Electro-Mechanics Co., Ltd., Suwon, Korea.
Corresponding e-mail: [Link]@[Link]
We present SAW filters based on a double resonance appearing in structures
consisting of long interdigital transducers (IDT) separated by a λ/4-gap. This
type of device was proposed by Plessky et al in 1996 [1]; it was called
degenerated coupled resonator filter (CRF) because its topology combines
characteristics of CRF and impedance elements. Although COM modeling
showed filter response with low loss and small standing wave ratio, experimental
results [1] gave unsatisfactory performance with 6-8 dB insertion loss. We show
that the loss was caused by scattering and attenuation in the gap, prominent in
this type of device with resonant energy concentrated in the gap. Replacing the
gap with a distributed gap will radically decrease the loss level [2]. To achieve
low loss, the anti-resonance frequency of the synchronous resonance of the
long IDTs must coincide with the frequency of the hiccup resonance arising in
the gap. The anti-resonance of the IDTs cancels the static capacitance of the
structure, self-matching the filter. We describe three variants: a single-ended 2-
IDT filter, a 3-IDT filter with balanced output, and a 3-IDT filter with cascaded
characteristics.
Topologically, the structures studied here resemble a standard CRF. The
operation, however, is not similar to that of a CRF. Long structures give rise to
strong reflections that prevent acoustic propagation in the device. In a 2-IDT
device, electrical signal is transmitted from input to output by excitation of a
synchronous resonance in the first IDT and a hiccup resonance in the gap. Gap
resonance transmits a part of the acoustic energy to the second IDT, creating a
voltage in the bus bars and a synchronous resonance in the IDT. The hiccup
resonance is spatially localized in the gap, as opposed to CRF where resonance
modes are distributed over the length of the device. Therefore, we feel that it is
more precise to call the structure a double-resonance filter.
Above-described scheme of operation can be applied to situation with more
than one gap; e.g., for 3 IDTs separated by 2 distributed gaps, transmission of
signal from input at first IDT via center IDT to output at third IDT is possible, if
the central transducer is left open-circuited. The characteristics of such a filter
are similar to that of two 1-gap filters cascaded. Grounding the center IDT
eliminates the uniform resonance in this part of the structure and prevents the
transfer of the signal from input to output.

45
The 3-IDT structure can operate with input in the center and output at sides,
typical of a standard CRF. In this case, the resonances in the two gaps are
connected in parallel, and the skirts will not be as steep as for the cascaded
variant. Such a symmetric structure can be employed with unbalanced input
and balanced output.
Measurements confirm the simulations of the described scheme of operation,
and experimental data will be presented. The structures described here are
ideal for filters with 1...1.5 % bandwidth requiring extremely low loss and small
standing wave ratio.
[1] V. P. Plessky et al, 1996 IEEE Ultrasonics Symposium, pp. 25-28.
[2] W. Wang et al, 2004 IEEE Ultrasonics Symposium, pp. 1363-1366.

PS-14
IDENTIFICATION OF NEW LTO HVPSAW
ORIENTATIONS CONSIDERING FINITE
THICKNESS ELECTRODES.
T. KENNY* and M. PEREIRA DA CUNHA, University of Maine, Orono.
Corresponding e-mail: mdacunha@[Link]
The continuing trend towards higher operating frequencies in RF communications
has strained SAW filter fabrication capabilities, focusing recent efforts on
identifying low-attenuated high velocity pseudo-surface acoustic wave
(HVPSAW) orientations of quartz, lithium niobate (LNO), lithium tantalate (LTO),
and lithium tetraborate (LBO). In particular, LTO is a well established SAW
substrate widely used for wireless applications, due to its relatively high
electromechanical coupling and moderate temperature sensitivity in comparison
to the other substrates previously mentioned. Although HVPSAW propagation
directions have been recently identified in the literature for these materials,
several show diminished coupling or prohibitively high propagation losses. Others
imply the use of heavy mechanical loading by the use of thick electrodes to
achieve reduced device loss.
This paper reports on the investigation of LTO HVPSAW excitation and
propagation properties, including: phase velocity, propagation loss per
wavelength (λ), and electromechanical coupling (K2) for arbitrary orientations
under periodic aluminum (Al), gold (Au), and platinum (Pt) electrodes. The
harmonic admittance of periodic electrodes is calculated using orthogonal
Chebyshev basis functions in conjunction with the finite element method. The
determination of well defined complex poles in the harmonic admittance curve
as a function of crystal rotation leads to the identification of low-attenuated and
high coupled HVPSAW orientations. For example, along selected orientations
of the LTO (0°,120°,ψ) plane, HVPSAWs are identified which exhibit phase
velocities around 5000 m/s, propagation losses as low as 0.04 dB/λ, and K2=0.74
%, for Al and Au electrodes with thicknesses h/λ=9% and h/λ=3%, respectively.
For the same electrode materials and thicknesses, the HVPSAWs propagating
under a uniform film layer show propagation losses around 0.3 dB/λ, using matrix
method calculations. Therefore, the resulting decrease in propagation loss by
more than one order of magnitude demonstrates the importance of the grating

46
structure in achieving low loss operational HVPSAW devices. The reported work
identifies new LTO HVPSAW orientations, exhibiting high phase velocity, low
propagation loss, and high electromechanical coupling, suitable for high
frequency, low loss communication and sensor applications.
This work was conducted with support from the National Science Foundation
(grants ECS-0233463 and ECS-0134335).

PS-15
HEXAGONAL SAW DEVICES
FOR ENHANCED SENSING.
S. CULAR*1, V. BHETHANABOTLA1, and D. BRANCH2, 1University of South
Florida, Tampa, FL, 2Sandia National Laboratories, Albuquerque, NM.
Corresponding e-mail: cular@[Link]
We present the design, fabrication and testing of a novel, hexagonal array based
on 36° YX lithium tantalate for non destructive evaluation of thin inorganic and
biological films. Propagation along the x-axis generates the leaky shear horizontal
(SH) mode where off axis propagation excites Rayleigh type surface acoustic
waves (SAW). Our approach permits rapid and simultaneous extraction of
multiple film parameters. Given that SAW devices are in widespread use for
chemical and biological sensing; a need exists to rapidly interrogate the interface
for several parameters. This is especially relevant in biological applications when
the sample quantity can be very limited. Our design allows for the simultaneous
extraction of multiple properties (film material density or thickness, Lamé and
shear moduli, sheet conductivity) of a thin film material to achieve a more
complete characterization than when a single SAW device is utilized. In sensor
applications, this capability translates to better discrimination of the analyte and
possibly more accurate determination of the concentration. We present this
device as an alternative to a SAW sensor array configuration that does not
allow for simultaneous probing and which would require larger devices and
packaging. These cost advantages are significant when working with novel
piezoelectric materials. The device is based on a double split finger delay-line
with widths of 4 µm and a delay path of 197λ. The individual delay path of each
hexagonal device intersects in the center of the die producing a single region
for sensor analysis. Additionally, the central region where the acoustic waves
intersect is shorted to reduce the number of modes of waves traversing the
surface. Initial testing has shown the pass band frequency of the individual delay
paths to be centered around 97 MHz. The acoustic velocities of the rotated
device have been measured to be 3452 m/s, 3161 m/s, and 3331 m/s, which
correspond to the theoretical range of all acoustic waves in the crystal of velocities
3290-4160 m/s. This variation is sufficient to allow for the simultaneous solution
of the same wave parameter dependent equation that will allow for the multiple
properties of the sensing film to be extracted.
We would like to thank Nancy Saldanha and Dr. Don Malocha of the University
of Central Florida for the use of fabrication equipment and insightful discussions.
Support for this work has been provided by NSF grant number DGE-0221681
and the University of South Florida CERC.

47
PS-16
A HIGH FREQUENCY VARIABLE FOCUS ULTRASONIC
TRANSDUCER USING POLYUREA THIN FILM.
M. NAKAZAWA*1, T. KOSUGI1, K. NAKAMURA1, S. UEHA1, A. MAEZAWA2,
and Y. HIRAO3, 1Tokyo Institute of Technology, Yokohama, Japan, 2Konica Minolta,
Tokyo, Japan, 3Kobayashi Institute of Physical Research, Tokyo, Japan.
Corresponding e-mail: nakazawa@[Link]
Polyurea thin film has useful characteristics as a high frequency ultrasonic
transducer material such as a low acoustic impedance of seventy percentages
of that of PVDF, considerably high piezoelectric constants compatible with
conventional piezoelectric polymers, and a higher Curie temperature. Fabrication
on curved surfaces and multi-layer lamination are possible by vapor deposition
polymerization. However, few studies have been made on the application for
ultrasonic transducers.
This report presents fabrications and experimental evaluations of the non-
focusing, the line/point-focusing, and the variable focus transducers with polyurea
thin films. First, the vapor deposition polymerization process using two monomers
is briefly reviewed, and the temperature condition for higher piezoelectric
constants is explored. Flat polyurea thin film transducers of about 1 µm thickness
are deposited on polyimide films of 25 µm, while focusing transducers are
fabricated on spherical plano-concave optical lenses with diameters of 20 mm
and radius curvatures of 25 mm, 50 mm, 80 mm, and 100 mm. Here, the polyurea
layer of 1 - 100 µm is sandwiched by two aluminum electrodes.
The resonant frequency, the mechanical Q, and the electromechanical coupling
factor were evaluated through the electrical measurement with an impedance
analyzer. Pulse transmitting/receiving experiments were carried out at the
resonant frequencies to determine focusing parameters. These experiments
show that the polyurea transducer works effectively at over 100 MHz and is
suitable for a high frequency diagnostic ultrasound transducer.
A line focusing transducer with the variable focal length was also prototyped. A
polyurea thin film of 5 - 7 µm is deposited on a flexible polymer film as a transducer
element. The film transducer is attached in the gap between two mechanical
sliders, so that the transducer may be bent and its radius of curvature may be
mechanically controlled by the gap width. The continuous change of the focal
length from 0.8 mm to 2.54 mm was successfully demonstrated in the pulse
transmitting/receiving experiments.
A high resolution B-mode imaging of living tissue using the polyurea focusing
transducer will be discussed as a preliminary study for fabrication high density
array transducers.

48
PS-17
3-D ULTRASOUND IMAGING USING FORWARD-
VIEWING CMUT RING ARRAYS FOR INTRAVASCULAR
AND INTRACARDIAC APPLICATIONS.
D. YEH*1, O. ORALKAN1, I. WYGANT1, M. O’DONNELL2, and B. KHURI-YAKUB1,
1
Stanford University, Stanford, CA, 2University of Michigan, Ann Arbor.
Corresponding e-mail: dtyeh@[Link]
Forward-viewing intravascular ultrasound enables new procedures in medicine
such as diagnosing severely occluded blood vessels or guiding the placement
of stents. Using a forward-viewing transducer in catheter-based applications
requires clearance in the center of the transducer for the guidewire, and requires
the development of a ring array. It is challenging to design and fabricate
piezoelectric transducers of the dimensions needed for forward-looking ring
arrays. However, Capacitive Micromachined Ultrasonic Transducers (CMUTs)
can be used to develop such arrays. This paper presents the first full synthetic
phased array volumetric images from CMUT ring arrays with integrated
electronics.
CMUTs offer several advantages over piezoelectric transducers for use in medical
imaging. Microlithography is used to make CMUTs and routinely yields the finely
controlled dimensions required for high frequency ring arrays. In addition, the
CMUT can be switched on-demand between its two modes of operation during
the imaging procedure, thereby providing a choice between the low frequency
conventional mode for navigation and the high frequency collapse mode for
diagnosis.
The CMUT ring array demonstrated in this paper has the following parameters:
ring diameter, 2 mm; number of elements, 64; element pitch, 102 µm; element
size, 100×100 µm; cells per element, 9; cell membrane radius, 13 µm.
The imaging setup consists of the CMUT mounted in a 209-pin PGA package
with the array elements wire bonded to 64 transmit/receive channels in a bank
of four integrated circuits. The full synthetic phased array data set was collected
by exciting one element at a time, and acquiring signals from the entire array for
each transmit element. Each A-scan was acquired at a sampling rate of 500
MS/s for both collapse and conventional modes, with 16 averages, and without
averaging. The imaging phantom consists of three steel wires 0.30 mm in
diameter. A-scan data indicates that the transducer operates conventionally at
7.5 MHz with a 6-dB bandwidth of 8.5 MHz, and in collapse mode at 19 MHz
with a bandwidth of 13 MHz. A conical volume was reconstructed offline from
the complete 64×64 set of A-scans, weighted for full-aperture resolution and
cosine apodization. The wires are clearly visible in images reconstructed from
data taken with no averaging, and demonstrate the good SNR of the CMUT.
Because of higher operating frequency and reduced crosstalk, collapse mode
produces images with a narrower main lobe than conventional mode imaging.
We have demonstrated 3-D ultrasound imaging with a forward-viewing CMUT
ring array using the first full synthetic phased array data taken using a ring
array. A fully integrated system with flip-chip bonded electronics will further

49
improve performance and SNR. These results show that forward-looking probes
for intravascular ultrasound are well on their way to becoming reality.
This work is funded by the National Institutes of Health. David Yeh is supported
by a National Defense Science and Engineering Graduate Fellowship.

PS-18
ANALYSIS AND DESIGN OF DUAL-ELECTRODE
CAPACITIVE MICROMACHINED
ULTRASONIC TRANSDUCERS.
R. O. GULDIKEN* and F. L. DEGERTEKIN, Georgia Institite of Technology,
Atlanta.
Corresponding e-mail: gtg181u@[Link]
Although ultrasound imaging arrays based on conventional Capacitive
Micromachined Ultrasonic Transducers (CMUTs) have been shown to have
adequate performance, the flexibility offered by microfabrication techniques can
be exploited to further enhance the capabilities of CMUT arrays. With multiple
electrodes embedded in dielectric CMUT membrane, dual-electrode CMUTs
that can be considered an example of this approach improving both transmit
and receive performance of the CMUTs without relying on collapsed membrane
operation [1]. Simple parallel-plate electrical equivalent circuit model suggests
that the electromechanical transformer ratio close to collapse depends only on
the gap thickness but not on the absolute value of the collapse voltage. The
advantage of the dual-electrode structure in the receive mode becomes apparent
when non-zero DC bias is applied to the side electrodes reducing the effective
gap for the center receive electrode. However, in contrast to the ideal parallel
plate CMUT, the maximum improvement with gap thickness reduction is limited
in CMUTs with dielectric membranes. As predicted by coupled finite element
analysis, when the bias voltage applied to the side electrodes is increased over
a certain value, the bottom of the silicon nitride membrane underneath the center
electrode makes contact with the bottom surface before the device is forced
into unstable collapse region and reach the maximum possible receive sensitivity.
Therefore, minimizing the dielectric layer thickness between the receive electrode
and the bottom electrode, e.g. eliminating the isolation layer over the bottom
electrode, helps improve the receive performance. These predictions are verified
by experimental data obtained on dual-electrode CMUTs by recording receive
signal amplitude as a function of side and center electrode bias voltage. Dual-
electrode CMUT design for transmit performance improvement involves
placement of side electrodes and shaping the silicon nitride membrane to
increase maximum volume displacement. For example, using 0.4µm deep, 1µm
wide notches between 4µm wide center and side electrodes in a 0.9µm thick
20µm wide rectangular silicon nitride membrane improves the overall
performance of CMUT by 10.7dB over conventional CMUT. This is achieved
while reducing the required maximum (collapse) voltage levels from 90V to 50V
for center receive electrode when 70V is applied to the side electrodes. Analysis
methods, experimental results, and optimal dual-electrode CMUT designs are
discussed in detail in this paper.

50
[1] R.O. Guldiken, J. McLean and F.L. Degertekin, “CMUTs with Dual-electrode
Structure for Improved Transmit and Receive Performance,” submitted to IEEE
Trans. Ultrason., Ferroelect., Freq. Contr., 2005.

Session: P1A
BLOOD FLOW
Chair: E. Ebbni
University of Minnesota
P1A-1
SYNTHETIC APERTURE FLOW ANGLE ESTIMATION
ON IN-VIVO DATA FROM THE CAROTID ARTERY.
N. ODDERSHEDE* and J. A. JENSEN, Center for Fast Ultrasound Imaging,
Technical University of Denmark, Denmark.
Corresponding e-mail: no@[Link]
In conventional ultrasound velocity estimation systems only the velocity projected
onto the direction of the steered ultrasound beam is found. It has previously
been shown how true blood flow velocity magnitudes can be found using synthetic
transmit aperture imaging. The method is based on cross-correlation between
lines beamformed along the flow direction. This method assumes the direction
of flow is known.
Jensen (2004) presented a method for estimating the direction of flow. The
angle determination method is based on a search for the maximum normalized
cross-correlation as a function of angle. It assumes the largest correlation is
seen for the angle of flow. Previously, this method has only been validated using
data from a circulating flow rig. This paper presents an In-Vivo investigation of
the method.
Real time data covering a full cardiac cycle of the carotid artery of a healthy 34-
year old male volunteer is acquired and then processed off-line using a large
computer cluster.
Data are acquired using our RASMUS experimental ultrasound scanner and a
128 element 7 MHz linear array transducer with λ pitch. A 20 µs chirp was used
during emission and virtual transmit sources were created in front of the
transducer using 64 transmitting elements. Data from 9 transmissions with each
64 receiving elements are beamformed and coherently summed to create high
resolution lines at different angles for a set of points within the region of flow.
The pulse repetition frequency was set to 16.5 kHz.
The direction of flow is estimated using the above mentioned method. It is
compared to the flow angle of 110° with respect to the axial direction, determined
visually from the B-mode image. For a point in the center of the common carotid
artery, the flow angle is estimated with a bias of -2.2° and a standard deviation
of 20.5° over time.
Blood velocity magnitudes along the individual angle estimates are found and
compared to velocities estimated along the visually determined direction. The

51
velocity difference had a bias of 0.0014 m/s and a standard deviation of 0.14 m/
s. The peak systolic velocity was around 1 m/s.
Full color flow maps from different parts of the cardiac cycle are presented,
including vector arrows indicating both estimated flow direction and velocity
magnitude.
This work was supported by grant 26-04-0024 from the Danish Science
Foundation and by B-K Medical A/S.

P1A-2
REAL-TIME ADAPTIVE CLUTTER REJECTION
FILTERING IN COLOR FLOW IMAGING USING POWER
METHOD ITERATIONS.
L. LOEVSTAKKEN*1, S. BJAERUM2, K. KRISTOFFERSEN2, and H. TORP1,
1
Norwegian University of Science and Technology, Trondheim, Norway, 2GE
Vingmed Ultrasound, Horten, Norway.
Corresponding e-mail: [Link]@[Link]
The removal of signal from stationary and moving tissue is a necessity and a
challenge in ultrasound Color Flow Imaging (CFI) and related techniques. This
task has conventionally been performed by using fixed response temporal high-
pass filters prior to blood flow parameter estimation. By adapting the clutter
rejection filter to the clutter signal statistics, better attenuation of the clutter signal
can be achieved in normal as well as more extreme cases of clutter movement
and acceleration. This improvement can provide a better detection of blood flow
and ensure less biased velocity estimates, which in the clinical setting can
translate to a more clear diagnosis of patients with abnormal flow patterns related
to pathology.
We propose a new algorithm for real-time adaptive clutter rejection filtering in
CFI related techniques, based on the eigenvector decomposition of the signal
correlation matrix in image regions of uniform clutter statistics. The filtering is
performed by projecting the signal vector for a given sample volume onto the
complement of an eigenvector basis representing the clutter signal subspace.
This method has previously been shown to be superior to conventional fixed
response high-pass filtering, but has until now been considered too
computationally demanding for real-time processing in general higher frame
rate modalities.
By using packet acquisition with interleaving, the acquisition time is short between
neighbouring beams, and sampling of uniform clutter statistics by averaging in
the lateral as well as the radial direction of the image can be properly obtained.
A rectangular averaging grid coinciding with the interleave group size determines
the number of filters that need to be estimated for each frame. Using the iterative
power method technique, the dominant eigenvalues and corresponding
eigenvectors within each averaging region can effectively be estimated. The
power method converges rapidly when the dominant eigenvalue is large relative
the remaining eigenvalues. This is shown to be the case when clutter is present,
and by using the convergence rate of the power method as a criterion for deciding
if there is remaining clutter, a clutter subspace basis can be found.

52
The real-time performance of the algorithm is sufficient on standard CPUs used
in desktop computers today, and a filter prototype has been implemented on a
GE Vingmed Vivid 7 ultrasound system. By using the number of algorithmic
floating point instructions (flops) as a measure of performance, the algorithm
has been compared to that of FIR and polynomial regression filtering. For a
representative example, the flop count for filtering one packet frame using the
new algorithm was 2.7 Mflops, compared to 1.2 Mflops for polynomial regression
filtering, and 0.50 Mflops for FIR filtering. Images of real-time scanning using
the new algorithm will be given, showing the superiority of adaptive filtering for
detection of blood flow and for removing clutter artifacts.

P1A-3
A NEW EIGEN-BASED FLOW ESTIMATOR USING THE
MATRIX PENCIL APPROACH.
A. YU*1, L. MO2, R. WARRINER1, and R. COBBOLD1, 1University of Toronto,
Toronto, ON, Canada, 2ZONARE Medical Systems, Mountain View, CA.
Corresponding e-mail: cobbold@[Link]
In color flow imaging, velocity estimation is often performed by applying a clutter
filter to remove low-frequency components in the slow-time signal and then
using a correlation-based strategy to find the mean flow velocity. However, the
clutter filter may distort the flow spectrum and lead to significant estimation
biases when the flow spectrum is near the filter cutoff frequency. As such, our
aim is to develop a more consistent flow estimation strategy that performs well
in various flow regimes.
In this paper, we describe a new eigen-based flow estimation strategy that can
be directly applied to the slow-time signal without prior clutter filtering. Our new
estimator, which is based on a framework originally developed for system pole
estimation, works by exploiting the algebraic properties of a mathematical entity
called matrix pencil and in turn treating flow estimation as a generalized
eigenvalue problem. Similar to a previously reported eigen-based estimator [1]
known as MUSIC, the Matrix Pencil can be considered as a principal spectral
component estimator whose eigenstructure is assumed to be the set of dominant
complex exponentials in the slow-time signal. Hence, by including the clutter
harmonics as part of its eigenstructure, this new estimator can extract the mode
flow velocity in the presence of clutter. In comparison to MUSIC, the Matrix
Pencil should give more consistent flow estimates since it does not rely on
covariance matrix statistics. Also, because its formulation does not involve
eigenroot searches after eigendecomposition, it should be more computationally
efficient.
We studied the Matrix Pencil estimator using raw color flow data acquired with
a 5 MHz linear-array scanner and a steady flow phantom (5 mm tube diameter).
For this study, the clutter-to-blood signal ratio near the vessel wall was estimated
to be between 20-30 dB. As well, the Matrix Pencil assumed that the slow-time
signal has two principal complex exponentials: one corresponding to the clutter
harmonic and the other to the flow harmonic. For comparison, flow estimation
was separately performed using an autocorrelator with a Chebyshev-I high-
pass filter (5th order, step initialized, and normalized cutoff frequency of 0.1).

53
Our results indicate that the estimates from the Matrix Pencil were comparable
to the ones obtained from the autocorrelator. In particular, given a 14-point
ensemble length and a center line velocity of 70 cm/s, the average estimation
difference between the two approaches was only 2.4 cm/s. Also, our spectral
analysis of individual color flow pixels show that the two mode frequency
estimates from the Matrix Pencil were close in location with the two largest
Fourier coefficients of the slow-time power spectrum.
We anticipate that the Matrix Pencil can be applied to various flow scenarios
such as the microcirculation. Indeed, this new estimator should be particularly
useful in micro-flow studies because the presence of tissue motion near the
microcirculation would considerably increase the complexity of clutter
suppression.
Reference: [1] Vaitkus and Cobbold, IEEE Trans. UFFC, pp. 939-954, 1998.

P1A-4
DETERMINATION OF VELOCITY VECTOR
ANGLES FOR THE DIRECTIONAL
CROSS-CORRELATION METHOD.
J. KORTBEK*1,2 and J. JENSEN1, 1Center for Fast Ultrasound Imaging, Technical
University of Denmark, Denmark, 2B-K Medical, Herlev, Denmark.
Corresponding e-mail: jk@[Link]
For conventional blood velocity estimation systems the velocity estimate is a
projection of the true velocity onto the ultrasound beam. It is thus impossible to
detect transverse flow, and these systems rely on knowledge of the angle between
the flow vector and the direction of the ultrasound beam, which traditionally is
determined visually from B-mode image.
In this paper a method is suggested for determining both velocity magnitude
and angle in any direction in the image plane without using B-mode image and
with a computational demand which permits implementation in a conventional
scanner. Beamforming along a given direction creates spatial signals denoted
directional signals. These signals are focused along the velocity direction, and
a cross-correlation is used for finding the correct velocity magnitude. The angle
is found from beamforming directional signals in a number of directions and
then selecting the flow angle as the direction with the highest normalized
correlation value between consecutive directional signals.
The focusing of directional signals is a key element in this method. Different
ways of calculating the time-of-flight and the influence on the estimation results
are presented. The standard deviation for the angle estimation for a flow angle
of 60° improves from 4.5° to 2.8°, when the transmit time-of-flight is calculated
using a virtual source instead of between transmit aperture center and focusing
point.
The approach is investigated using Field II simulations and data from the
experimental ultrasound scanner, RASMUS. A 7 MHz linear array transducer is
used with a normal transmission of a focused ultrasound field. The performance
of the velocity and the angle estimator is presented for parabolic flow at angles

54
45°, 60°, 75° and 90° with a peak velocity of 0.3 m/s. Not more than 20 pulse-
echo emissions are used for the estimations.
For the velocity magnitude estimation, a mean relative bias and standard
deviation are calculated over the profile. In simulations the bias and standard
deviation for the velocity magnitude are all 1% for flow at 45° and 60°. At 75°
and 90°, the bias is 4% in both cases and the standard deviations are 2% and
7%, respectively. For the angle estimation, a probability of having estimates
within the range ± 15° from the true angle is calculated. The bias and standard
deviation of the angle estimation are calculated from the estimates that are
within that range. In simulations the bias and standard deviation are between
0.3° and 2.3° and the probabilities are above 99%. Measurements are made
with the experimental scanner RASMUS, and for velocity estimation the standard
deviation is between 2.5% and 14% for flow between 45° and 90°. For angle
estimation the bias and standard deviation at 60° to 90° are between 1.0° and
3.4° and the probabilities are above 98%.
This work was supported by grant 9700883, 9700563 and 26-04-0024 from the
Danish Science Foundation and by B-K Medical.

P1A-5
BLOOD VELOCITY ESTIMATION USING SPATIO-
TEMPORAL ENCODING BASED ON A FREQUENCY
DIVISION APPROACH.
F. GRAN*, S. NIKOLOV, and J. A. JENSEN, Center for Fast Ultrasound Imaging,
Technical University of Denmark, Copenhagen, Denmark.
Corresponding e-mail: fg@[Link]
This paper investigates the possibility for flow estimation using spatio-temporal
encoding of the transmissions based on a frequency division approach in
synthetic transmit aperture imaging (STA).
In STA, a major disadvantage is that only a single transmitter (denoting single
transducer element or a virtual source) is used in every transmission. The
transmitted acoustic energy will be low compared to a conventional focused
transmission where a large part of the aperture is used. By using several
transmitters simultaneously the transmitted energy can be increased.
However, to focus the data properly, the information originating from the different
transmitters must be separated. To do so, the pass band of the transducer is
divided into a number of sub-bands with disjoint spectral support. At every
transmission, each transmitter is assigned one of the sub-bands. In receive, the
signals are separated using a simple filtering operation. To attain high axial
resolution, broad-band spectra must be synthesized for each of the transmitters.
By multiplexing the different waveforms on different transmitters over a number
of transmissions this can be accomplished. To further increase the transmitted
energy, the waveforms are designed as linear frequency modulated signals
implying that the full excitation amplitude can be utilized during most of the
transmission.

55
The method has been evaluated for blood velocity estimation for several different
velocities and incident angles. The program Field II was used. A 128 element
transducer with a center frequency of 7 MHz was simulated. 64 transmitting
element were used as the transmitting aperture and 128 elements were used
as the receiving aperture. Four virtual sources were created in every transmission.
Each virtual source was created by 16 transducer elements focused 5 mm in
front of the transducer. To generate a smooth broad-band spectrum for each
virtual source, eight excitation waveforms had to be designed. Every transmission
yields a temporally narrow-band low resolution image for every virtual source.
By multiplexing the waveforms for the different virtual sources and combining
the result of several transmissions, broad band excitation waveforms are
synthesized for each virtual source. By beamforming lines in the flow direction,
directional data was extracted and correlated. Hereby, the velocity of the blood
was estimated. The pulse repetition frequency was 16 kHz.
Three different setups were investigated with flow angles of 45, 60 and 75 degrees
with respect to the acoustic axis. Four different velocities were simulated for
each angle at 0.10, 0.25, 0.50 and 1.00 m/s. The mean relative bias with respect
to the peak flow for the three angles was 2.34, 2.61 and 4.85% and the mean
relative standard deviation with respect to the peak flow was 0.98, 0.86 and
1.27% respectively.
This work was supported by grant 9700883, 9700563 and 26-01-0178 from the
Danish Science Foundation and by B-K Medical A/S, Denmark.

P1A-6
DESTRUCTION-MODE OPTOACOUSTIC FLOW
MEASUREMENTS WITH GOLD NANORODS.
P.-C. LI*1, S.-W. HUANG1, C.-W. WEI1, and C. R. WANG2, 1Department of
Electrical Engineering, National Taiwan University, Taipei, Taiwan, ROC,
2
Department of Chemistry and Biochemistry, National Chung Cheng University,
Chia-Yi, Taiwan, ROC.
Corresponding e-mail: paichi@[Link]
Time-intensity based blood flow measurements have been widely applied in
various imaging modalities. In optoacoustic imaging, we have previously tested
the hypothesis that gold nanospheres can be used for quantitative flow estimation
based on wash-out analysis. In this study, a wash-in flow analysis method using
gold nanorods is developed. Specifically, the proposed technique makes use of
the shape dependence of the optical absorption wavelength of gold nanorods
and their shape transformation after pulsed laser irradiation. Gold nanorods
have unique optical properties based on their surface plasma absorption and
the transformation results in a shift in the peak absorption wavelength. The
shape transformation property does not exist on gold nanospheres. Also, the
photon-induced shape transformation of the gold nanorod system exhibits mainly
a rod-to-sphere conversion. Therefore, with pulsed laser irradiation at a
wavelength that corresponds to the peak optical absorption of the original shape
of the gold nanorods, the particles that undergo shape changes can be viewed
as being destructed by laser irradiation at that wavelength. A series of laser
pulses with lower laser energy can then be applied to monitor the acoustic

56
intensities as a function of time as new gold nanorods wash in the region of
interest. Quantitative flow information can then be derived from the temporal
intensity data. Based on this property, a destruction-mode technique was
developed. With this technique, a destruction laser pulse is followed by a series
of observation pulses. Phantom experiments were conducted. 0.26 nM of gold
nanorods were injected into a vessel with a diameter of 5 mm connected to a
polyvinyl alcohol phantom. The gold nanorods had an optical absorption peak
at 1018 nm. A Nd:YAG laser operating at 1064 nm with a pulse duration of 5 ns
was used for optical illumination. The laser radiation power was 73.1 mW for
the destruction pulse and 11.3 mW for the observation pulse. The pulse repetition
rate was 15 Hz. An ultrasonic single element transducer with a center frequency
of 1 MHz was employed to detect the acoustic signals. After detection, the
acoustic waveforms were amplified and sampled at 100 MHz by a data acquisition
card housed in a personal computer on which subsequent data analysis was
performed. Laser irradiation was from the top of the vessel and the acoustic
detection was from the side. To prevent recirculation of the nanoparticles back
in the phantom, the water only passed through the phantom once. The flow
velocity ranged from 0.1 mm/s to 2.3 mm/s. The linear regression results show
that the correlation coefficient between the measured velocities and the true
values was close to unity (0.96), thus demonstrating the feasibility of the proposed
destruction-mode technique for quantitative flow estimation in optoacoustic
imaging.

P1A-7
A SIMPLE METHOD TO REDUCE ALIASING
ARTIFACTS IN COLOR FLOW MODE IMAGING.
J. UDESEN*1,2, S. I. NIKOLOV1, and J. A. JENSEN1, 1Center for Fast Ultrasound
Imaging, Technical University of Denmark, Lyngby, Denmark., 2B-K Medical A/
S,, Herlev, Denmark.
Corresponding e-mail: ju@[Link]
It is a well known limitation in conventional blood velocity estimation using a
phase estimation approach, that aliasing artifacts are present, when the blood
velocities exceed a value determined by half of the pulse repetition frequency
(the Nyquist frequency) of the ultrasound system. In this paper we propose a
simple method to increase the aliasing limit to twice the Nyquist frequency.
The method is based on a conventional autocorrelation estimator. Normally
N=8 to N=16 emissions in the same direction are used to estimate the axial
velocity. In the suggested method N/2 emissions are fired at a pulse repetition
period Tprf. After a time period of 1.5 Tprf the N/2 emissions are repeated. All N
emissions are used to determine the mean phase vector P1 for emissions
separated Tprf in time. To determine whether aliasing is present, another phase
vector P2 is estimated for the two emissions N/2 and N/2+1. The sign of the
angle θ between P1 and P2 is used to determine whether aliasing occurs. θ is
found by estimating the angle to the vector P = P1* P2, where (*) denotes complex
conjugate. The method is limited to angles where -π<θ<π. This corresponds to
Doppler frequencies smaller than twice the Nyquist frequency 1/(2Tprf) of the
system. To increase signal to noise ratio on the estimates, RF-averaging is
performed over a pulse length in the axial direction.
57
The method is tested in simulations using the Field II program. A virtual 7 MHz
linear array transducer is used to scan a virtual blood vessel of radius 6.7 mm,
situated at a depth of 40 mm at an angle of 70 deg to the ultrasound beam.
Hanning apodization is applied both in transmit and receive. The pulse repetition
time Tprf of the system is 0.5 msec, and the scatters are moved, so that parabolic
flow is simulated with a maximum velocity of 0.4 m/s. These values give aliasing
at the center of the vessels inside a radius of 3.15 mm. The velocity is found
along one axial line where N=8 emissions are used for each velocity estimate,
and 24 velocity profiles are found along the axial line. The proposed method
gives a mean standard deviation and a mean bias of the axial velocity over the
vessel of 1.3%, and 3.5%, respectively. This should be compared to the case
where a conventional autocorrelation estimator is used to find the axial velocity.
Here the mean bias over the vessel is 23.3%, and the mean standard deviation
is 5.3%. The large bias of 23.3% is due to the aliasing.
The method is tested experimentally using the ultrasound scanner RASMUS
and a circulating flowrig with parabolic flow. All the setup parameters are kept
the same as in the Field II simulation. When the proposed method is used, the
mean bias over the vessel is 1.2%, and the mean standard deviation is 4.8%.
This should be compared to the conventional case, where the mean bias over
the vessel is 20.0%, and the mean standard deviation is 3.9%.
This work was supported by grant 9700883,9700563 and 26-01-0178 from the
Danish Science Foundation, the Ministry of Science, Technology and
Development, and by B-K Medical A/S, Denmark.

Session: P1B
ELASTOGRAPHY
Chair: P.-C. Li
National Taiwain University
P1B-1
ANALYSIS OF TRANSIENT SHEAR WAVE
GENERATION FOR REAL-TIME SHEAR
ELASTOGRAPHY IN-VIVO.
S. YAGI*1, A. SANUGA1, Y. KONDO2, K. TAMURA2, and M. SATO3, 1Meisei
University, Oume, Tokyo, Japan, 2Aloka Co., Ltd., Oume, Tokyo, Japan,
3
Microsonic Co., Ltd., Tokyo, Japan.
Corresponding e-mail: yagi@[Link]
We developed a numerical analysis of transient shear wave generation and
experimentally verified the results using our customly modified real-time
displacement tomography for real-time imaging of dynamic absolute shear
modulus in vivo.
FDTD analysis: For real-time evaluation of the transient shear wave generation
Finite-Difference Time-Domain analysis of elastic wave field is utilized, in which
the separation of compressional and shear waves, dynamical analysis of shear

58
wave across the near-field and the inversion approaches to absolute shear
modulus can be implemented. According to the compressional and shear wave
velocities in liver tissue (1500 m/s and 1 m/s), 2-dimensional spatial mesh size
and time step are set to 1mm × 1mm and 0.33 ms respectively. The FDTD
analysis is performed through the 150mm × 150mm area surrounded by perfect
matching layer to absorb reflected waves. Since the spatio-temporal impulse
response by vertical stress on the tissue surface clearly demonstrates the split
bi-directional shear wave profiles separated by vector velocity potentional, as
the far field directivity by Miller and Pursey in1954, non-splitting transient profile
by the vertical stress distribution are evaluated by adjusting band-limited temporal
driving function. After the successive experimental analysis it is confirmed that
the vertical stress distribution by superposing each separated reverse-polar
raised-cosine can generate a continuous shear wave front in circular phase
after the end of temporal driving by raised-cosine. The spatial separation and
width of raised-cosine and the applied driving time are focused on 10mm, 20mm
and 50ms in this tissue model, respectively.
Experiments: To verify the performance of transient shear wave generation a
conventional ultrasonic tomography was modified to real-time imaging of
longitudinal displacement between successive RF echo frames (12-bit. 4 times
the center-frequency of transducer). These spatio-temporal data are processed
by the parallel architecture of flexible Field Programmable Gate Array modules
for real-time local cross-correlation and its statistical phase. After the scan-line
conversion the real-time tissue movement is superposed upon B-mode image
at the same frame rate. The measuring accuracy was verified by the shear
wave profile across a liver tissue in vivo, in which the displacement by heartbeat
was less than 10 µm per frame interval and the boundary traveled at 1 m/s or
so. The experimental stress pulse on the surface of tissue phantom with 1 to 5
% gelatin concentration clearly demonstrated the predicted shear wave
generation and propagation with velocity of 0.5 to 3 m/s across the phantom
medium.
This is partly supported by JHPF research grant 2004.

P1B-2
A PARAMETRIC STUDY ON PROCESSING
PARAMETERS FOR TWO-DIMENSIONAL CARDIAC
STRAIN ESTIMATION: AN IN-VIVO STUDY.
S. LANGELAND*, P. F. WOUTERS, H. A. LEATHER, P. CLAUS, B. BIJNENS,
and J. D’HOOGE, Catholic University of Leuven, Leuven, Belgium.
Corresponding e-mail: [Link]@[Link]
In the last years, two-dimensional (2D) myocardial strain estimation has been
evolving as a method that could replace todays 1D strain estimation and solve
several of its limitations. The 2D strain estimation is typically based on 2D
displacement tracking of spatial features in the gray-scale (GS) or radio-frequency
(RF) data using either 1D or 2D kernels. All approaches have specific advantages
and disadvantages but a trade-off between accuracy and computational cost
generally has to be made. Although some studies have been presented in the

59
literature on the effect of these parameters, these have been limited to simulated
or tissue mimicking phantom setups.
The aim of this study was therefore to evaluate the influence of these parameters
on the strain estimate in an in-vivo setting in order to optimize of computation
time.
Methods: B-mode IQ data sets were acquired from the inferolateral wall of 4
open chest sheep using a parasternal long axis view. The data were aquired at
4.5 MHz at a sampling rate of 10 MHz using a Toshiba
PowerVision 6000 equipped with an RF interface. The framerate was 168 Hz
with a sector angle of 45 degrees and 60 image lines. The images were
interpolated by a factor of 10 in the azimuth direction. 2D displacements were
calulated in each data set based on a sum of squared differences estimator
using different parameters. The investigated parameters were: axial and azimuth
kernel size (2 and 3 wave lengths and 1, 3 and 5 beams respectively), axial
sampling frequency (40 and 80 MHz) and the tracking was performed both on
GS and RF images. Myocardial radial (R) and longitudinal (L) strain were
simultaneously estimated using a previously described methodology based on
2D motion gradients. Four segment-length sonomicrometry crystals (SM) were
placed in a tetrahedral configuration just lateral to the imaging plane giving a
continuous reference for the L and R strain. As an error measure, the mean
absolute difference between ultrasound and SM strain was calulated over the
cardiac cycle for all data sets and all processing parameters. Differences in
average error (DAE) between methods were statistically tested using ANOVA.
Results: No statistically significant differences were found for the different kernel
sizes in neither axial or azimuth direction (mean DAE 1D vs. 2D kernel = 0.1%,
p=NS). Similarly, neither sampling frequency (mean DAE = 0.4%, p=NS) nor
GS vs. RF tracking (mean DAE = 0.4% ,p=NS) had a significant influence on
the mean error. However, for all parameters L strain errors were significantly
smaller than the R ones (mean DAE = 3% ,p < 0.01).
Conclusion: Although it has been shown in the literature that under ideal
conditions the variance in the strain estimator reduces using a 2D kernel, these
results did not reproduce in this in-vivo setting. This might be due to an increased
sensitivity of 2D kernels to myocardial shear. Similarly, sampling frequency and
GS vs. RF tracking had no significant impact on accuracy. 1D kernels might
thus be better for 2D strain estimation in vivo since they give a similar accuracy
at a reduced computional cost.

P1B-3
ELASTOGRAPHIC PARAMETERS BY SURFACE
WAVE ANALYSIS.
N. BENECH, C. A. NEGREIRA*, and I. NUÑEZ, Facultad de Ciencias,
Montevideo. Uruguay.
Corresponding e-mail: nbenech@[Link]
Ultrasonic elastography has proved to be a useful means to obtain quantitative
information about the viscoelastic parameters of soft tissue. This information is
used to diagnose pathologies like cirrhosis or carcinomas surrounded of healthy

60
tissue in the breast. The analysis of surface waves in an elastographic experience
can be very useful and complementary in the diagnosis carried out starting from
the shear waves in the bulk of the medium. Previous work in this area is limited
although it is potentially useful as a diagnostic [Link] fact is due that surface
waves, as the shear waves, are very dependent of the viscoelastic parameters
of the medium. In this work we make a study of surface waves using the optic
method of desfocusing in a homogeneous phantom. We have made a sweeping
in frequencies from 40 up to 500 Hz, obtaining by this way a dispersion curve in
function of the medium’s viscosity. We discriminated the Rayleigh waves from
the Lamb waves by the speed dependance of the frequency. The speed of the
surface wave and the attenuation are compared with those obtained with transient
elastography from the shear waves in the interior of the phantom. For speed
values in the 1-5 m/s range we obtained a 2% agreement for this comparison. In
a second stage, we studied an heterogeneous phantom (a hard inclusion
surrounded of soft tissue). The lesion is detected through the amplitude variations
of the surface wave and this result was compared with elastographic image
made from the local shear wave speed. In this study we also determine the
exploration depth that this method allows in function of the excitation frequency.
This work was supported by Programa para el Desarrollo de las Ciencias Básicas
(PEDECIBA-URUGUAY) and project PROCISUR.

P1B-4
MEASUREMENT SPATIAL DISTRIBUTION OF HEART
WALL MOTION GENERATED BY REMOTE
PERTURBATION OF INNER PRESSURE.
H. KANAI*, H. HASEGAWA, and K. IMAMURA, Tohoku University, Sendai, Japan.
Corresponding e-mail: hkanai@[Link]
Background: Though the myocardial movability is important to diagnosis of the
heart diseases, a noninvasive method for estimation of the movability has never
been developed. If the heart wall is sinusoidally actuated using a low frequency
actuator directly attached to the chest wall, the bending vibration (mode-2)
dominantly occurs in the heart wall. For the vibration with mode-2, however,
there are 8 nodes which remain still on the heart wall. Thus, it is necessary to
distinguish the nodes from the other points to avoid underestimating the
movability of the regional myocardium. Method: By attaching an actuator to the
brachium artery and driving it by sinusoidal wave of f0 Hz, the inner pressure of
the artery is perturbed and the perturbation propagates along the artery to the
left ventricle (LV) of the heart. Then, the perturbation of the LV inner pressure is
generated. Using ultrasound-based method (Kanai et al. IEEE Trans UFFC 43,
1996), the resultant minute motion on the heart wall can be measured. Since
the vibration mode of the heart wall depend on the actuated frequency, the
vibration mode and the positions of the nodes can be identified from the
measurement of the spatial distribution of the heart wall motions by scanning
the ultrasonic beam (Kanai et al. Ultrasound Med Biol 27, 2001). Finally, from
the resultant strain and the delay of the strain to the applied pressure, the
instantaneous myocardial movability and its transition property during one cardiac
cycle are noninvasively estimated. Basic experiments: As a model of the heart,

61
a spherical shell (outer diameter=45 mm, thickness=7.5 mm) made of silicone
rubber was set in a water tank and a silicone rubber tube (50 cm in length) was
connected to the spherical shell. By remotely actuating the tube with a sinusoidal
vibration of f0 Hz, the internal pressure of the shell was perturbed with several
mmHg. From the measured spatial distribution of the vibrations on the shell
surface, the vibration-mode was identified. For 7-8 Hz, the vibration-mode is 0,
which shows to the homogeneous expansion and contraction without node. For
12 Hz, the mode is 1, which shows the parallel displacement of the whole shell.
For 16-17 Hz, the mode is 2. From its distribution, the positions of the nodes
were identified. In vivo experiments: By applying external remote actuation to
the brachial artery, the LV internal pressure was successfully perturbed for the
first time. The motions in and on the heart wall were measured with ultrasound
from the chest wall. For the actuation frequency of 26 Hz, the minute velocity
components of 0.01 m/s due to the perturbation of the inner pressure were
superposed on the spontaneous myocardial motion of about 0.1 m/s.
Conclusion: This study proposes a novel method to noninvasively perturb the
LV internal pressure by remotely actuating the brachium artery with the sinusoidal
vibration. By measuring the spatial distribution of the heart wall motion using
ultrasound, the vibration mode is identified, which has a potential to noninvasively
evaluate the movability and its transition property during one cardiac cycle.

P1B-5
TRANSIENT ULTRASOUND ELASTOGRAPHY FOR
BREAST CANCER DIAGNOSIS USING IMPULSIVE
RADIATION FORCE: AN IN VITRO STUDY.
D. MELODELIMA*1, J. BAMBER1, F. DUCK2, and J. SHIPLEY2, 1Royal Marsden
NHS trust and Institute of Cancer Research, Sutton, Surrey, UK, 2Royal United
Hospital MHS trust, Bath, UK.
Corresponding e-mail: [Link]@[Link]
Alternative imaging methods are needed to improve the effectiveness of breast
cancer detection and diagnosis. As a result, the development of imaging systems
capable of evaluating the mechanical properties of tissues with high resolution
is being carried out by numerous research teams. Several groups are studying
acoustic radiation force-based imaging modalities. However, to date these have
been largely based on the detection, by various means, of signals that are
proportional to the displacement induced by radiation force, whereas in pseudo-
static elastography it is generally strain that is imaged, as a surrogate for inverse
stiffness. The imaging of strain generated by radiation force from a focused
ultrasound transducer has never been suggested, and may be of interest to
improve breast cancer detection.
In this study, a focused ultrasound transducer of f-number 1.3 was driven at an
operating frequency of 1.7 MHz, and was used to apply localised radiation force
to a small volume of tissue mimic. Images were created using a single “pushing”
burst at each location, of duration 8 ms. A linear array with a centre frequency of
7.5 MHz was used to obtain echo signals. The focused transducer and the linear
array were aligned such that the focus was in the plane of imaging. The time-
dependent transient strains resulting from the impulsive radiation force were

62
mapped using a least-squares strain estimator and ultrasound RF correlation-
based displacement tracking methods.
Gelatine phantoms containing cylindrical stiff inclusions were used. Experimental
results demonstrate that strain on the order of 0.04% can be generated and
detected with a resolution of about 1 mm. A stiffer region exhibits lower strain
than a more compliant region. The instantaneous strain immediately following
cessation of the radiation force varies approximately linearly with the Young’s
modulus of the material. Good agreement was obtained between the size of the
inclusion seen macroscopically and measurements made using transient
radiation force elastography.
Strain imaging using impulsive radiation force offers several advantages. The
highly localised and transient strain that is produced by focused and impulsive
radiation force may permit the sensing of changes in tissue elastic properties
that are difficult to detect with conventional elastography, due to greater
independence from boundary conditions and a resulting improved global contrast-
to-noise ratio. For example, the characteristic, bi-directional, high strain artefacts
due to stress concentration, often seen with static elastography at the tissue-
inclusion interface, do not appear using the transient radiation force strain imaging
technique. It is also possible to utilize a single pushing burst, strongly focused
in two dimensions, that acts as a spatially localised impulse, and to use low
frequencies to supply radiation force more efficiently to deeper-lying tissues.
This work was supported by funding from the UK Department of Health (NEAT
D008).

P1B-6
THE EFFECT OF SURROUNDING GELATIN ON
ULTRASOUND GENERATED SHORT PULSE WAVE
PROPAGATION IN ARTERIES.
X. ZHANG* and J. GREENLEAF, Mayo Clinic College of Medicine, Rochester,
MN.
Corresponding e-mail: [Link]@[Link]
Background: Pulse wave velocity (PWV) is widely used for estimating the stiffness
of an artery. It is well known that a stiffened artery can be associated with various
diseases and with aging. Usually, PWV is measured using the “foot-to-foot”
method. The “foot” of the pressure wave is not clear due to reflected waves and
blood noise. Also, PWV is an average indicator of artery stiffness between the
two measuring points, and therefore does not identify local stiffness variations.
In a series of recent studies, we propose producing a bending wave in the arterial
wall using low frequency localized ultrasound radiation force and measuring the
wave velocity along the arterial wall [X. Zhang et al., Proc. 2003 Ultrason. Symp.,
1883-1886], [X. Zhang et al., Proc. 2004 SPIE, Medical Imaging]. The wave
velocity can be measured accurately over a few millimeters.
Objective: PWV is directly related to the Young’s modulus in the circumferential
direction of the artery by the well known Moens-Korteweg equation. However, in
this famous equation, PWV only relates to the artery properties and blood density,

63
but does not relate to the surrounding tissue. The objective of this paper is to
develop method for studying the surrounding tissue effect on the arterial waves.
Method: An excised artery was tested in a water tank. The artery was pressurized
internally with saline. A short pulse wave in the artery was generated by
ultrasound. The wave velocity was measured with laser. Then the artery was
embedded in a tissue-mimicking gelatin. The wave velocity of the artery
embedded in gelatin was measured.
Results: The short pulse wave velocity of the artery without gelatin is measured
5.6 m/s, while the short pulse wave velocity of the artery with gelatin is measured
10 m/s.
Conclusion: The short pulse wave generated by ultrasound in the artery travels
faster if the artery is embedded in gelatin. This means that the gelatin stiffens
the artery. Further research will be developing method for studying the effect of
surrounding tissue on artery stiffness.
This study is supported by grant EB 02640 from the National Institutes of Health.
The authors thank R. R. Kinnick for experimental support.

Session: P1C
TISSUE CHARACTERIZATION
Chair: E. Biagi
University of Florence
P1C-1
BREAST TUMOR CLASSIFICATION BASED ON IMAGE
SEQUENCE ANALYSIS DURING COMPRESSION.
K.-H. LEE1, Y.-H. CHOU2, C.-M. CHEN3, and P.-C. LI*1, 1Department of Electrical
Engineering, National Taiwan University, Taipei, Taiwan, ROC, 2Veterans General
Hospital, Taipei, Taiwan, ROC, 3Institute of Biomedical Engineering, National
Taiwan University, Taipei, Taiwan, ROC.
Corresponding e-mail: paichi@[Link]
Tumor contour features have been used for classification of breast tumors.
Typically such computer aided diagnosis systems are based on analysis of a
single image. In this paper, we investigate performance of breast tumor
classification using image features from a breast image sequence acquired when
the breast is under increasing compression. The motivation is to incorporate
features that are related to tissue elastic properties into classification. Also,
these features can be combined with the original contour features from a single
image to further improve the accuracy. Specifically, the features used in this
study included (1) area changing rate, (2) changing rate of the equivalent ellipse
parameters (long axis dimension, short axis dimension and inclination angle),
and (3) the mean displacement of the contour. The B-spline model was employed
to define the contour. By using this model, the analysis can be done efficiently
and classification can be performed in real time. 20 clinical image sequences
were obtained from the Taipei Veterans General Hospital and separated into

64
two categories by an experienced physician. Among the 20 cases, 7 were
classified as malignant nodules and the others were classified as benign nodules.
A linear support vector machine with the leave-one-out method was used for
training and classification. A success rate of 17/20 (85%) was achieved. The
method can be combined with existing single image contour features to further
improve the accuracy. In addition, it is a robust method because it is not sensitive
to the detailed structures of the contour. Finally, since only B-mode images are
needed, the proposed method can be more easily integrated with current clinical
systems than other elasticity imaging methods that require access to the radio
frequency data.

P1C-2
ULTRASOUND BASED ASSESSMENT OF
GEOMETRICAL PARAMETERS INVOLVED IN LUMEN
PRESERVATION DURING ATHEROSCLEROTIC
PLAQUE BUILD UP.
J. WENTZEL, A. DHARAMPAL, F. GIJSEN*, J. SCHUURBIERS, G.
RODRIGUEZ-GRANILLO, A. VAN DER STEEN, P. DE FEYTER, and C.
SLAGER, Erasmus MC, Rotterdam, the Netherlands.
Corresponding e-mail: [Link]@[Link]
BACKGROUND The presence of atherosclerotic plaques in human coronary
arteries is mostly identified by X-ray based angiography. However, during the
late eighties it became clear that the true atherosclerotic plaque burden is not
reflected in the dimensions of the lumen because of plaque compensation
mechanisms of the vessel wall, i.e. vascular remodeling. At the time, intravascular
ultrasound (IVUS) was the only tool to assess the true plaque burden of the
atherosclerotic plaques taking the total wall thickness into consideration. IVUS
studies showed that atherosclerotic plaques are often eccentric, implying that
the wall, including the endothelium, is only partially diseased. Mechanisms
responsible for remodeling of the vessel wall are thought to be dependent on
healthy endothelium. Therefore, we hypothesize that the plaque free wall (PFW),
having functional endothelium, controls the local arterial capacity of vascular
remodeling. We studied by IVUS whether reduction in PFW, presumable because
of plaque extension at the shoulders, attributes to lumen narrowing.
METHODS Coronary arteries of 5 patients (<50% angiographic stenosis) were
studied at baseline and at 4 years follow up applying an ECG gated stepwise
IVUS pullback (30 Mhz, CVIS, Boston Scientific). IVUS images were matched
carefully using anatomic landmarks. Only cross sections in between side
branches were studied. From the IVUS images the lumen and external elastic
membrane contours were semi automatically delineated to assess local wall
thickness (WT), lumen area (LA), lumen diameter, media bounded area (MBA)
and plaque area (PA) of successive cross sections at 0.5 mm intervals. WT
smaller than 0.2*lumen diameter was defined as normal and was used to
determine the per cross section the arc of PFW. Cross sections showing increase
in plaque area were selected and divided into 2 groups based on the median of
the change in PFW(∆PFW).

65
RESULTS In total 202 out of 336 cross sections showed increase in plaque
area over the 4 years follow up. The average increase in plaque area was 26%.
The median of the ∆PFW was -27.5°. Cross sections showing decrease in PFW
(-103.2°±$68.8°, p<0.05, n =101), presented with 15% lumen narrowing (p<0.05)
and still a minor increase in MBA of 2.5% (p<0.05). For cross sections showing
some extension in PFW (11.5°±32.7°, p<0.05, n=101) LA was preserved (+1.5%,
p=NS) and MBA increased (7.8%, p<0.05) more than the group with loss of
PFW. For both groups progression in plaque area was similar.
CONCLUSION IVUS derived plaque progression accompanied with decrease
in plaque free vessel wall, presumably because of plaque fissuring at the
shoulders, is related to lumen narrowing at 4 years follow up. These results
imply a potential role for the plaque free vessel wall in the process of lumen
preservation.

P1C-3
A NOVEL METHOD FOR AUTOMATIC CONTOUR
EXTRACTION OF ULTRASONIC BREAST LESIONS.
C.-K. YEH* 1, Y.-S. CHEN 1, and W.-S. CHEN 2, 1Department of Electrical
Engineering, Yuan Ze University, Taiwan, Chung-Li, Taiwan, 2Department of
Physical Medicine and Rehabilitation, National Taiwan University Hospital,
Taiwan, Taipei, Taiwan.
Corresponding e-mail: yehck@[Link]
Automatically detecting tumors and extracting lesion boundaries in ultrasound
images is difficult due to the variance in shape and the interference from speckle
noise. Previous research works primarily focused on lesion classification with
manual delineation of the tumor boundaries. In this study, we present a novel
algorithm to automatically find lesion contours in ultrasonic breast images. With
inherent property of ultrasound image embedding speckle noise, the image can
be regarded as degraded paint character (DPC) image containing closure noise,
which is well known in perceptual organization of psychology. An effective scheme
of removing closure noise using iterative disk expansion method has been
addressed in our previous work and it has been successfully applied to many
applications. The contour extraction of ultrasonic breast lesions can be equivalent
to the removal of speckle noise. In our method, the brightness equalization and
adaptive thresholding schemes are first used for converting the original
ultrasound image into a binary image, which is similar to DPC image. Then,
applying disk expansion method to the binary image, we can obtain a significant
radius-based image where the radius for each pixel represents the corresponding
disk covering the specific object information. Based on the cumulative distribution
function analysis on the radius-based image, the significant lesion region can
be easily located. Finally, a signal transmission process is used for searching
the complete breast lesion region and thus the desired lesion contour can be
effectively and automatically extracted. In this study, simulations and clinical
images were performed to evaluate the performance of the proposed algorithm.
Several types of cysts with different contours and contrast resolutions images
were simulated with speckle characteristics. The results show that the mean
normalized true positive area overlap between simulated contour and contour

66
obtained by the proposed algorithm is 87.5%. There is a strong correlation
between experienced physicians manual and automated contour extraction in
clinical breast images (R=0.90). The algorithm is also able to simultaneously
contour multiple lesions in a single image. Comparison with conventional snake
contour extraction algorithm, the proposed algorithm does not position any initial
seed point within the lesion and thus it can be viewed as a fully automatic process.
Furthermore, due to running time for contour extraction in a single breast lesion
image was 0.06 s on a 3.0-GHz Intel Xeon machine, the algorithm is feasible to
be implemented in a real-time ultrasound imaging system.
The authors acknowledge the National Science Council of ROC (NSC 93-2218-
E-155-002-) for providing support.

P1C-4
EVALUATION OF A MODIFIED AUTOCORRELATION
METHOD WHEN APPLIED TO CARDIAC STRAIN
RATE IMAGING.
A. BLOMBERG* 1, A. HEIMDAL 2, S. I. RABBEN 2, J. D’HOOGE 3, and A.
AUSTENG4, 1University of Oslo, Oslo, Norway, 2GE Vingmed Ultrasound, Oslo,
Norway, 3University Hospital Gasthuisberg, Leuven, Belgium, 4University of Oslo,
Oslo, Norway.
Corresponding e-mail: [Link]@[Link]
Background: Strain rate imaging is an ultrasound method suitable for quantitative
assessment of regional myocardial function. More precisely, strain rate measures
the rate of local deformation within a defined region of interest. Myocardial strain
rate imaging involves computing spatial gradients in tissue velocities. Since the
gradient-operation is noise-sensitive, accurate velocity estimates are essential.
When estimating tissue velocities using the conventional autocorrelation method
(AM), the ultrasound center frequency is assumed constant and equal to the
demodulation frequency, while in the modified autocorrelation method (MAM)
the former is continuously estimated together with the phase shift from pulse to
pulse. The AM is unbiased only if the demodulation frequency is equal to the
received center frequency. This assumption is not necessarily valid.
The MAM has shown promising results with respect to estimator variance and
bias when implemented in simulated environments.
Aim: The purpose of this work has been to evaluate the performance of the
MAM compared to the AM, when applied to myocardial strain rate estimation.
Method and results: Both methods have been implemented and evaluated in
two different environments, namely simulated RF-data from an analytic model
of the human heart and experimental data acquired from human hearts scanned
with a specially programmed Vivid 7 scanner (GE Vingmed Ultrasound).
When applied to simulated RF-data, the MAM demonstrated superior
performance with respect to estimator variance (43%-81% improvement
compared to the AM) and bias, as well as an over all slightly higher degree of
correlation with the true strain rate of the analytic model (r=0.98-0.99 vs. 0.97-
0.99, p<0.0001).

67
When estimating strain rate using experimental data, a previously presented
spectral strain rate method was used as a reference. In this case the MAM did
not show superior performance. On the contrary, the AM demonstrated a lower
estimator variance (by 46% -49%) and bias, as well as an over all higher degree
of correlation with the mean value of the strain rate spectrogram (r = 0.73-0.90
vs. 0.62-0.86, p<0.001).
There are several possible contributing factors to the degraded performance of
the MAM when applied to experimental data, including spatial scatterer
distribution and acoustic noise. In an attempt to explain the degraded
performance of the MAM, filtered white noise modeling reverberation noise was
added to the simulated RF-data before processing. SNRs ranging from infinity
(no added noise) to 5 dB were simulated.
Discussion: As expected, both methods displayed degraded performance in
the presence of noise. Interestingly, the MAM degraded more quickly. As the
SNR decreases, the performance of the MAM approaches that of the AM. For
an SNR of 5 dB, both methods display similar results.
In conclusion, the MAM shows promising results in simulated environments with
high SNR. Under heavy noise conditions and in experimental data, however, it
displays similar or poorer performance compared to the AM.

Session: P1D
BEAMFORMER IMPLEMENTATION
Chair: J.-Y. Lu
University of Toledo
P1D-1
“SYNTHETIC AXIAL ACQUISITION” FULL
RESOLUTION C-SCAN ULTRASONIC IMAGING.
Y. LI*, T. BLALOCK, W. WALKER, and J. HOSSACK, University of Virginia,
Charlottesville, VA.
Corresponding e-mail: jh7fj@[Link]
Previously, we described various components for a low cost approach for enabling
C-Scan beamforming using a fully sampled 2D array and very short element
data records. In the Directly Sampled In-phase and Quadrature (DSIQ) approach,
we use as few as one complex sample per array element. However, there is a
penalty in terms of reduced final image spatial and contrast resolution when
using DSIQ as opposed to conventional techniques using finite length time
records and delay / sum beamforming. In the ‘Synthetic Axis Acquisition’ (SAA)
approach, we take advantage of the fact that we use an unfocused transmit
beam (plane wave) and are imaging primarily shallow and slowly moving, or
even static, tissue regions. This allows us to acquire, over a set of approximately
25 serial transmit / receive firings, each with the receive sampling trigger delay
offset by the equivalent of one digital sampling interval, a per channel data
record of sufficient length to enable an approximation to conventional full record
delay and sum beamforming. The penalty for taking this approach is encountered

68
in several areas. Firstly, the opportunity to increase SNR by averaging is reduced.
Secondly, the performance of the approach degrades as a function of tissue
motion. Additionally, it is readily evident that this version of a C-Mode imaging
device requires more extensive hardware and signal processing. However,
depending on choice of beamforming condition (center frequency, aperture size
and focal depth), the SAA approach yields an improvement in lateral resolution
by a factor of approximately two over that achievable with DSIQ. FIELD II based
simulations were initially performed for the parameters appropriate for our next
generation of ‘Sonic Window’ that we intend to present in another paper at this
conference. These parameters are: 5 MHz Center Frequency, 60 × 60 fully
sampled 2D array with 0.3 mm pitch. The imaging depth was 20 mm and the
assumed pulse repetition interval was 40 microseconds. Assuming a required
final equivalent sampling rate of 40 MHz necessitated 25 Synthetic Axial
Acquisitions (‘pulse-echos’) to form sufficiently long data records. Our numeric
simulations have been performed for a variety of target velocities occurring during
the process of acquiring the full data record. It has been observed lateral target
velocities of 50 mm/s or more are tolerable with no significant (i.e. greater than
1-2 dB at approximately the -20 dB level) impact on beamplot performance. As
might be anticipated, the approach is more sensitive to degradation due to axial
target motion. However, even in this case an axial velocity of 50 mm/s has an
impact of approximately 3-4 dB at the -20 dB level. A variety of both simulation-
based and experimental data results will be presented. Further improvements
based on using aperiodic pulsing as an approach to enable faster acquisition
are also discussed.
NIH EB002349, EB001826

P1D-2
A HARDWARE EFFICIENT BEAMFORMER FOR SMALL
SIZE ULTRASOUND SCANNERS.
J. Y. LEE*, H. S. KIM, and T. K. SONG, Sogang University, Seoul, South Korea.
Corresponding e-mail: powerjun77@[Link]
Digital receive beamformer is one of the most important parts governing the
performance and complexity of a medical ultrasound imaging system. In this
presentation, we present a hardware efficient digital beamformer for small size
ultrasound scanners, which is advantageous over the two most widely used
beamforming schemes, interpolation beamformer and phase rotation
beamformer, in terms of hardware complexity and performance. In the
interpolation beamformer, the rf data are sampled at a rate not smaller than 4fo
and the sample rate is increased by a factor of 4 to achieve the required time
delay resolution of 1/16fo, which requires a multi-tap interpolation filter for each
channel. In the phase rotation beamformer, the inter-channel delay compensation
is done by shifting the phase of inphase and quadrature components of the
input rf data, which only requires 4multipliers. However, the phase rotation
beamformer requires an I/Q demodulator for each channel, consisting of two
mixers and two multi-tap LPFs. Moreover, accurate delay compensation can be
done only at the center frequency.

69
The proposed beamformer is identical to the interpolation beamformer except
that it uses fractional delay(FD) filters to generate the delayed samples, instead
of the interpolation filters. To obtain the 1/16f0 delay resolution, each FD filter
should be capable of providing a programmable fractional delay D, which has
the ideal response H(w)=exp(-jwD), D=0,0.25,0,5,0.75, and |H(w)|=1, for all w.
A 4tap FD filter was designed using maximally flat filter design constraints and
least square error method. It was shown that its phase response error increases
with frequency: For a sampling rate of 40MHz, the group delay error is 0.001 at
3.5MHz, 0.0015 at 5MHz, 0.0208 at 7.5MHz, and 0.0756 at 10MHz. It was also
shown that the magnitude response error, which also increases with frequency,
has less effect on the accuracy of the delayed samples compared with the phase
response error.
To evaluate the proposed method, computer simulations were performed with
Gaussian signals of 50% 6dB bandwidth and the signal to noise ratios(SNR),
defined as the ratio of true sample value to the difference between the true and
delayed sample values, of different schemes were compared. The SNR of the
proposed FD beamformer decreases with frequency (58dB at 3.5MHz, 55dB at
5MHz, 37dB at 7MHz and 22dB at 10MHz), whereas the phase rotation
beamformer has SNR lower than 20dB at all frequencies. In addition, the phase
rotation beamformer requires 36multipliers: 2 for mixers, 32 for two LPFs and 4
for phase rotation. It was also observed that the SNR of the interpolation
beamformer is lower than that of the FD beamformer by 20dB at 5MHz and 6dB
at 10MHz, when 4multiplier polyphase filter is used to implement a 16tap
interpolation filter. In order for the interpolation beamformer to have the same
SNR as the FD beamformer, its hardware complexity should increase by from
25% to 300% depending on the imaging frequency. These results show that the
FD beamformer outperforms the phase rotation beamformer and interpolation
beamformer in terms of both performance and hardware complexity.

P1D-3
MODERN IMPLEMENTATION OF A REALTIME 3D
BEAMFORMER AND SCAN CONVERTER SYSTEM.
K. WALL* and G. LOCKWOOD, Queen’s University, Kingston, Ontario, Canada.
Corresponding e-mail: kwall@[Link]
We have developed a scalable high-speed beamformer and scan converter for
real-time 3D imaging. The beamformer uses 4 Field Programmable Gate Array
(FPGA) devices (Xilinx Spartan3 series), each chip processing the data received
from 72 of the 288 transducer elements. Internally, the FPGAs are divided into
a series of element processing units. Each unit handles the data processing
from a single transducer element, storing the received data in a circular buffer.
An internal calculator then determines the appropriate delay samples required
for each target point, pulls them from this buffer, interpolates and apodizes them.
The summed result is transferred to the PC for display. For a high-speed FPGA
based beamformer, two areas typically consume the largest amount of the
available resources: look-up table storage ROM, and resources used to switch
signals between the internal processing units. For our application, the
beamformer is required to generate a 60 degree sector format image. Storing

70
the focusing delays associated with the range of steering angles used to form
the sector format image, takes a huge amount of memory. An exponential
reduction in the size of this table is possible by modifying the beamformer to
generate scan lines that are perpendicular to the array in much the same way a
linear array image is formed. This greatly reduces the size of the look-up table
since the same set of delays can be shared for all the image lines. This choice
also allows a significant reduction in switching resources by allowing table data
to be passed between the internal data processing units, rather than distributed
from a single source. To further reduce the size of the stored data, compression
techniques are applied. Storing only the second order derivative of the data
allows a 10:1 reduction in storage. The beamformed data is transferred to a PC
through the PCI bus. A PCI bus controller integrated into the FPGA code provides
a DMA transfer into the computer’s RAM. When each transfer block is complete,
a PC side program coded in C# .NET begins a second DMA transfer onto the
video card, where all scan conversion logic is implemented. Offloading scan
conversion processing to the Graphics Processing Unit (GPU) provides a huge
increase in performance as the GPU is designed for handling massive parallel
data processing. It also leverages the built in graphics functions, such as
interpolation, anti-aliasing, and alpha blending. The scan conversion, log
compression, and display are all coded in a High Level Shader Language
program, using Pixel Shader 3.0. The FPGA design has been tested using the
Xilinx XST synthesis engine. Functional testing using Modelsim generates 100%
agreement with theoretical Matlab simulation results, and shows secondary lobe
suppression of at least 55 dB. The beamformer can sustain processing for 7,500
transmit pulses/sec, internally calculating more than 3,840 beamformed points
per pulse. Volume frame rates far exceeding 30fps are easily achievable by the
GPU scan converter, with very low CPU dependence.

P1D-4
ABOUT THE POSSIBILITY TO IMPLEMENT A
NONUNIFORM OVERSAMPLING RECEIVE
BEAMFORMER IN A FPGA.
L. LIE* and M. E. TANASE, Politehnica University of Timisoara, Timisoara, Timis,
Romania.
Corresponding e-mail: [Link]@[Link]
A hardware implementation is presented in a context of receive beamforming
based on nonuniform oversampling techniques. Using these techniques is
possible to obtain image quality similar to those obtained using multi-bit digital
beamforming techniques, in the conditions of significant complexity, size and
cost reduction. The beamformers based on delta-sigma modulators have two
major advantages over traditional multi-bit methods: simplifying the ADC structure
and ensure precise delays by manipulating the samples taken with a high rate.
The digital beamformers that are based on non-uniform sampling create the
dynamic focusing by sampling the echo samples at specific moments that ensure
the coherent summing operation at the reception.

71
From a structural point-of-view, the digital beamformer with a nonuniform
oversampling is made up of two sections:
- An analogue section containing the delta-sigma modulators.
- A digital section consisting of the following entities: Digital controller, FIFO,
Adder and decimating filter.
The central entity of our design is the digital controller, which generate the non-
uniform sampling clocks for command the delta-sigma modulators and load the
FIFO buffers, and ensures the signals for download the FIFO buffers and for
synchrony the adding block and the decimating filter.
In order to generate the non-uniform sampling clock we have taken into account
the possibility of iterative “in circuit” calculating for the delay information, based
on the “midpoint” algorithm applied to one of the equations that derives from the
focusing geometry.
The design has been simulated in the ALTERA QUARTUS environment and the
resulting structure of has been implemented in the largest FLEX10K circuit
EPF10K250ABC600. The device utilization shows that it can host the sampling
clock generators for an array with 64 elements. Therefore, it is possible to
implement in a single FPGA, which contains memory blocks, every digital blocks
of a beamformer: the digital controller and FIFO memory for each channel, the
adder for n (64, 128,...) channels and the decimating filter.
The beamformer was tested using ultrasonic data acquired with a multi-bit
converter. Using a MATLAB routine the original signal was converted into delta-
sigma format. The resulting signal was used like test signal for the FPGA
implemented beamformer. The filtered signal was captured in a MATLAB script
for comparative analyse in time and frequency domain with the original signal in
order to validate the implementation.
The results shows the possibility to build in the future a “single chip” receive
beamformer using programmable analogue / digital structures.

P1D-5
PHASED SUBARRAYS FOR LOW COST
C-SCAN APPLICATIONS.
Y. LI*, T. BLALOCK, W. WALKER, and J. HOSSACK, University of Virginia,
Charlottesville, VA.
Corresponding e-mail: jh7fj@[Link]
The performance of subarray beamformer processing as a means of reducing
overall beamforming computational load for a low cost C-Scan imaging device
is analyzed. Unlike the case encountered in B-Mode imaging, the C-Scan format
is characterized by a large channel count (64 × 64 = 4096 in this case) and very
short data record lengths. Additionally, unlike the case in B-Mode, there is no
essential need for realtime processing of ‘streaming’ received channel data.
Thus, the C-Scan image format benefits greatly from a subarray processing
approach in which partial sums are ‘reused’ with modified differential delays to
form multiple beamformed image pixels. It is shown that the beamforming
computational load for our previously described low-cost C-Scan imaging device

72
(‘Sonic Window’), measured in terms of delay and sum operations, may be
reduced by approximately 50 fold with a 1-3 dB impact on sidelobe level at the
-30dB level. Subarrays with dimensions of 2, 4, 8 and 16 elements per side
have been investigated. Larger subarrays requires fewer calculations but exhibit
higher sidelobes. Subarrays with side dimensions of 2 and 4 elements do not
have significant image performance advantages over a subarray size of 8 but
require more significantly more computational effort. Consequently, for the array
/ target geometries of interest to us (5 MHz and an aperture / focal depth
corresponding to f/number = 1), a subarray dimension of 8 × 8 is favored. The
impact of subarrays on performance along the diagonal directions (worst case)
and also the benefits of optimal (i.e. hexagonal) beam spatial sampling are also
considered. Additionally, results using both conventional full record length and
single complex sample (‘Directly Sampled In-phase and Quadrature’ - DSIQ)
are presented. The potential computational savings is less when using DSIQ
because imaging resolution is inferior and consequently relatively fewer image
samples are required to fully sample the image plane without aliasing. Initial
results have been obtained using FIELD II simulations. Experimental results,
obtained with our prototype ‘Sonic Window’, are also presented.
NIH EB002349, EB001826

P1D-6
AN ARBITRARY WAVEFORM TRANSMITTER USING
BIPOLAR PULSERS BASED ON A HIGH ORDER
MODIFIED SIGMA DELTA MODULATION.
H. H KIM*, H. S HAN, and T. K SONG, Sogang university, Seoul, South Korea.
Corresponding e-mail: kh2_brian@[Link]
Coded excitation has been studied to improve the SNR with the limited excitation
voltage, to increase the imaging frame rate, to improve the performance of
ultrasound imaging with contrast agents, and so forth. However, it requires a
complicated arbitrary waveform transmitter for each active channel that
customarily uses a multi-bit DAC and a linear power amplifier(LPA). Not only
does the LPA increase the cost and size of a transmitter block, but it consumes
much power, increasing the system complexity further and causing a heating-
up problem, which is critical in some cases.
We present an efficient arbitrary waveform generation scheme, which is based
on modified sigma-delta modulation(SDM) methods suitable for coded excitation
techniques relying on correlation based pulse compression. The proposed
arbitrary waveform transmitter is composed of a 1bit DAC to generate single-bit
SDM waveforms, a bipolar pulser to produce their high voltage versions, and a
simple LC filter connected to array elements, resulting in a great reduction in
the hardware size and cost. At the present work, the modified sigma-delta
modulator has been designed to produce a weighted chirp signal with a center
frequency of 5MHz and 6dB bandwidth of 3MHz. It was assumed that the Nyquist
rate for this chirp is 20MHz. Customary methods to obtain a high SQNR in
single-bit SDM are to use a high oversampling ratio(OSR) and to add dither
noise to the quantizer input. In ultrasound imaging, however, high OSR should

73
be avoided since for example an OSR of 16 requires high voltage pulsers that
should operate at switching rate as high as 320MHz. Our design goal is to
achieve a low OSR of 4 (i.e., 80MHz switching rate), while the peak sidelobe
level of the compressed waveform on receive should be smaller than -45dB. To
achieve such a low OSR, a modified 4th order SDM is employed, which also
eliminates the necessity of using dither noise. Its noise transfer function(NTF) is
chosen to obtain an optimal stability and to suppress the inband quantization
noise, using a linear coefficient comparison method and root locus method for
stability analysis. For this purpose, the two coefficient sets of the modified SDM
are selected such that the NTF has the poles and zeros of the 4th order
Butterworth and Chebyshev type 2 highpass filters, with the cutoff frequencies
of 8MHz and 6MHz, respectively.
To evaluate the proposed scheme, the 80MHz single-bit SDM waveform was
passed through a 4th order Butterworth LC lowpass filter with a cutoff frequency
of 7MHz, yielding the demodulated SDM signal with a SQNR of 25dB. The
demodulated signal was then correlated with the original weighted chirp and
compared with the ideally compressed signal, i.e., the autocorrelation of the
original chirp. The peak sidelobe of the compressed waveform in the proposed
method was -45.0dB, whereas the ideally compressed one exhibits the peak
sidelobe level of -46.5dB, both with the same mainlobe width. The results showed
that the proposed scheme can be used to reduce the hardware complexity of
the coded excitation imaging without sacrificing the image resolution.

Session: P1E
MEDICAL IMAGING
Chair: J d’Hooge
University of Leuven
P1E-1
PHASE BASED LIVER MOTION COMPENSATION
OF HARMONIC IMAGES.
A. KISSI* 1 , A. BOUAKAZ 1 , L. POURCELOT 1 , S. CORMIER 2 , and F.
TRANQUART1, 1INSERM Unity 619, Tours, France, 2LERI CRESTIC, Reims,
France.
Corresponding e-mail: kissi_a@[Link]
Background: The use of ultrasound contrast agents allows to image liver perfusion
in real time.
This visualization provides important information for the diagnosis of various
diseases as well as for characterization of tumors. By using the slopes of the
rise in activity during the arterial phase of liver enhancement, an perfusion index
value β is computed. This index allows us to characterize lesions. However,
artefacts, such as respiratory or cardiac motions often limits this quantification.
A preliminary motion compensation is thus needed.
Method: In this context, we present a new method for compensating motion in
order to estimate the perfusion index, over a sliding window, in order to accurately

74
characterize liver lesions from harmonic ultrasound sequences. The differential
motion estimation technique used in this study is the phase-based optical flow.
The local phase gradient, intensity-amplitude invariant, is first calculated by
using Gabor quadrature filters. From the output of these filters, the component
of lateral and transversal motion are computed by minimization procedure. This
components are thus used to compensate motion.
Results: This method has been tested off-line five sequences of 160 non
compressed images, using a Siemens Acuson Sequoia 512 with a 4C1 probe,
after Sonovue injection (Bracco SpA imaging, Italy). The study was in CPS mode
with an mechanical index of 0.21.
Our process eliminates some errors in positioning and thus in the calculated
intensity from a given lesion. Our results show that, for solitary metastasis, there
is an improvement of lesion quantification of 15%: the perfusion index β=0,21.
This study is funded by the International Rotary Club. This is gratefully
acknowledged.

P1E-2
SPECTROSCOPIC THREE-DIMENSIONAL IMAGING
OF LIGHT SCATTERING MEDIUM BY DETECTION
OF ULTRASONIC VELOCITY CHANGE DUE
TO LIGHT ILLUMINATION.
H. HORINAKA*1, T. URA1, Y. NAKATANI1, K. WADA1, and T. MATSUNAKA2,
1
Osaka Prefecture University, Sakai,Osaka,Japan, 2 Aloka Co. Ltd.,
Mitaka,Tokyo,Japan.
Corresponding e-mail: horinaka@[Link]
Optical tomography using near-infrared light is theoretically capable of imaging
not only soft tissue information but also metabolic information of a living body
using spectroscopic method. However, it was very difficult to construct the optical
image of deep region of the body because the near-infrared light is scattered
severely inside the soft tissue. We have already proposed a new optical
tomography for medical diagnosis based on the interaction between the light
and the ultrasonic wave. The 2D (two-dimensional) optical absorption image
was constructed by detecting the ultrasonic velocity change due to the thermal
agitation of light scattering medium. In this study, to apply diagnosis of
physiological activities and pathological defects we performed experiments to
construct the spectroscopic 3D (three-dimensional) optical images of the highly
scattering phantom by developing 2D optical image construction. An ultrasonic
transducer with the center frequency of 5 MHz and the beam width of 1.9mm in
water was used for the experiments. Two oval spheres (about 30x20x20mm)
which were made of agar included many small pieces of plastic thin film (0.1mm
in thickness and 3 to 5mm in width) were prepared as experimental phantoms:
one was transparent and the other was colored with the black ink. They were
mounted in a transparent container filled with aqueous solution of the 10%
“Intralipid” which has the scattering coefficient of 16cm-1 equivalent to the reported
value of biomedical tissue. The transducer emitted ultrasonic pulses in water
and received echo pulses from the phantom. The light from a laser diode (810nm)

75
were guided around the container by optical fibers and illuminated the phantoms
for thermal agitation. The waveforms of echo pulses with and without light
illumination were detected and time delay between two pulses were measured.
The ultrasonic velocity shift was detected under the illumination intensity of 0.2W/
cm2(the skin exposure limit of continuous wave diode laser in the ANSI Z 136.1
Safe Use of Lasers standard). The 2D images of the ultrasonic amplitude and
the velocity shift due to illumination were measured to construct the 3D images
when the transducer was shifted every 1mm. The ultrasonic amplitude 3D image
showed the contour of two oval spheres in the scattering medium. On the other
hand, the ultrasonic velocity 3D image showed only the oval sphere made of
colored agar. It was confirmed that the spatial information of optical absorption
was obtained in highly scattering medium by our method.

P1E-3
MULTITONE NONLINEAR CODING.
A. NOWICKI*, J. WÓJCIK, and W. SECOMSKI, Institute of Fundamental
Technological Research, Polish Academy of Sciences, Warsaw, Poland.
Corresponding e-mail: anowicki@[Link]
A new method termed multitone nonlinear coding that utilizes nonlinear properties
of acoustic tissue to improve ultrasound image resolution is presented. It is well
known that under nonlinear propagation conditions field distribution for
compressional and rarefactional portion of the wave is different. If P+(x,t) is a
solution of the linear propagation equation fulfilling the boundary condition
P+(x,t)=P+(S(x),t) for x at S(x) where S(x) is the surface of the source then P-
(x,t)=-P+(x,t) is also the solution of the same equation for boundary condition
with reverse phase P-(S(x),t) =-P+(S(x),t) and P+(x,t)+P-(x,t)= 0. For nonlinear
propagation the sum of P+ and P- is not equal to 0 despite the fact that, P+(S(x),t)+
P-(S(x),t)=0.
In our novel method the acoustic source is activated with two tones burst (2.5
and 5 MHz) with specially designed polarization of the adjacent tone burst. This
new approach is called nonlinear multi tone coding NMC because the choice of
polarization of the adjacent tones (and their amplitudes) allowing optimization
of the receiving properties depends on the nonlinear properties of the tissue.
The calculations were done for two tone burst propagating in the tissue-like
lossy medium comprising of spheres of different diameters and cylinders with
inhomogeneous acoustic impedance, average absorption 7 Np/m•MHz.
Comparison of the spatial field distribution obtained using conventional harmonic
imaging approach, in which 2-nd harmonic is used to reconstruct the image,
and field distributions corresponding to each of the excitation frequencies and
differential frequency are presented.
The received signal comprises the similar amount of harmonics as in the standard
pulse inversion method, however, in comparison with the last it has also
fundamental component enhancing the image content close and distant to the
transducer, where harmonics are limited.
Conclusions: In conventional harmonic imaging a single tone burst is applied to
the acoustic source and the 2-nd harmonic signal is utilized to generate the
resolution improved image. However, in the standard harmonic imaging only

76
half of the transducer bandwidth is employed, whereas in the proposed multitone
nonlinear coding full bandwidth of the ultrasound imaging transducer is utilized.
The results indicate that in comparison with conventional harmonic the acoustic
energy available for image construction when using multitone nonlinear coding
is enhanced by the factor of 2 with the corresponding increase in signal-to-
noise ratio.

P1E-4
ULTRASOUND BLADDER SCANNER BASED ON 2D
CONCAVE PROBE.
S. DERROUICH*, J. OYAMA, T. HIGUCHI, and T. ABE, Nagasaki University,
Nagasaki City, Japan.
Corresponding e-mail: derrouich_salah@[Link]
The medical reports in Japan show that around 1 of 5 senior citizens suffers
from urine incontinence. Although the humidity sensor installed inside the diapers
is a valuable measure for those patients, it is still a subjective system of only
reporting the fact that the diaper is wet. In the context of this background, there
is a need to develop a warning sensor system to evaluate the quantity of urine
while still inside the bladder.
The abdominal scans are performed using a sector probe in both longitudinal
and transverse planes. The bladder may also be seen using a trans-vaginal or
trans-rectal approach with an intracavitary probe; these probes have to be
attached to an ultrasound machine for visualization. Those types of scanners
are impractical for bladder volume estimation because they are large, heavy,
table top and expensive devices.
In this framework, the key bladder-scanner related challenges are the following:
periodic and automatic scanning, portable, small size, light and affordable.
The bladder scanner presented in this paper uses A-mode approach and a
limited number of UT; using A-mode approach with a small number of Ultrasonic
Transducers (UT) is the only reasonable way to deal with all the above mentioned
challenges. Our scanner is certainly not the first bladder scanner that makes
use of A-mode approach with a small number of UT. However the existing bladder-
scanner uses either one-dimension (1D) linear or circular array probe; these
probes can sometimes fail to detect the bladder efficiently, especially in case of
a small Pubis-Peritoneum-Window (PPW) or the absence of symmetry when
the bladder is shifted to the right or left side of the body.
Inspired by the Pinhole-Camera-Model, our probe has been geometrically
associated to constitute a 2D concave probe. The potency of this assembly
subsists in the arrangement of 12 UT in a 4x3 matrix, the UT are angled in a way
to face the center of the PPW. The probe increases scan rate, precision and
reliability, it is also capable of sensing even in both cases of small PPW and
bladder-asymmetry, and allowing different part of the bladder to be reached by
a maximum number of used transducers; then shape of the bladder is estimated
and approximated for volume calculation. By setting the system to scan
automatically and periodically, when the bladder reaches a preset threshold a
warning alarm is activated.

77
The probe has been designed, produced, tested and used with a developed
algorithm for experiments on human-like model filled with water, targeting a
bladder-like balloon. Experiments are conducted and results are compared with
‘Yuririn USH-052’ and ‘Aloka ProSound SSD 3500’ particularly in case of small
PPW and bladder-asymmetry.
The results show a better performance of our scanner in critical conditions. An
optimum size, weight and price are expected in the final product. Furthermore,
conducting experiments directly on human body is the framework of our future
research
This work is supported in part by the Cooperation of Innovative Technology and
Advanced Research in Evolution Area, Toshi Area Project.
This work is registered for Japanese Patent No. 2005-126558.

P1E-5
DESIGN AND EVALUATION OF A LOW COST PC CARD
USED FOR AN ACOUSTIC RADIATION FORCE BASED
MECHANICAL CHARACTERIZATION SYSTEM.
M. SANTY*, T. BLALOCK, and W. WALKER, University of Virginia, Charlottesville,
VA.
Corresponding e-mail: mks3k@[Link]
A growing area of ultrasound research is the use of acoustic radiation force to
characterize mechanical properties of soft tissues. In acoustic radiation force
imaging, the transfer of momentum related to the propagation of an acoustic
wave through a dissipative medium is used to generate small and localized
displacements within tissues. These displacements are then tracked using time
delay estimation techniques. Images of viscoelastic properties are formed by
processing temporal changes in displacement.
In this paper we describe the development of a new system that consists of a
custom printed circuit board housed in and controlled by a standard PC. All
transmit and receive electronics, prior to digitization, are placed on one PCI bus
compatible card. PCI bus interaction is performed by a PCI interface chip (ASIC
Design Services, PCILite chip, Midrand, South Africa). The front end electronics
consist of transmit and timing circuitry. The transmit circuitry is a bipolar transmit
circuit, composed of a cascade of paired N- and P-type mosfets, with
programmable on and off times operating at a maximum transmit voltage of
121V and up to frequencies of 25MHz. The timing circuitry, controlled by a
Complex Programmable Logic Device (CPLD) (Xilinx Inc., XCR3512XL), controls
the length of the pulse, the number of pulses and transmissions, and the pulse
repetition frequency. The receive electronics consist of two stages, a protection
and an amplification stage. The protection stage is composed of multiple diode
pairs configured to clip high voltages, protecting downstream electronics. The
amplification stage consists of two non-inverting opamp based amplifiers with a
total gain of 60dB. After the amplification stage, the data is sampled at 50MHz
using a PC card (Gage Applied, Compuscope 12100 A/D card, Lachine, Quebec,
Canada). This sampling is synchronized with the transmit pulse to assure timing
accuracy on the order of picoseconds. The digitized data is then processed

78
using MATLAB (The Mathworks, Inc. Natick, MA). We are currently developing
an easy to use interface for users who do not have extensive ultrasound expertise.
The PC card is fully assembled and being tested for functionality. All circuitry is
being characterized to ensure timing accuracy of transmit and receive signals.
We have also performed a series of experiments in a water tank. A half cycle of
a 60Vpp, 8.3MHz sinusoid was transmitted using a single piston transducer. A
2.2V echo was successfully received and digitized. Experimental data is in relative
agreement with simulation results performed in PSpice. A number of experiments
are currently being performed to evaluate the potential of using this custom
system to characterize tissue properties. We present experimental results and
displacement curves from water tank and various soft tissue experiments. This
system has the capability to transmit and receive acoustic pulses that can be
processed to determine mechanical data of various soft tissues.
We acknowledge support from the Whitaker Foundation.

P1E-6
HIGH-FREQUENCY HIGH FRAME RATE ULTRASOUND
IMAGING SYSTEM FOR SMALL ANIMAL IMAGING
WITH LINEAR ARRAYS.
X.-C. XU*, C.-H. HU, L. SUN, J. YEN, and K. SHUNG, University of Southern
California, Los Angeles,CA.
Corresponding e-mail: xiaochex@[Link]
Small animal imaging is needed by biological and pharmaceutical research.
Currently both Micro-CT and Micro-PET were developed and used. But their
cost effectiveness and real-time capability are still of a great concern. Especially
due to the fast heart rate of a mouse (>400/min), cardiovascular research utilizing
mice requires imaging modalities with high frame rate capability (>100 Frame/
sec), which can not be realized by Micro-CT and Micro-PET at present. High-
frequency ultrasound imaging on the other hand is capable of achieving a better
spatial resolution at an affordable price. Single element transducer based
ultrasound backscatter microscopies (UBM) have been used in small animal
research. Recently, high frequency ultrasonic arrays (>30MHz), which provide
clinical convenience, reduce imaging time, and offer dynamic focusing, have
been developed successfully. Comparing to current UBMs, a linear array based
high-frequency ultrasound imaging system would alleviate many of these
problems.
In this paper, we report the development of a high-frequency high frame rate
ultrasound imaging system using 30~35MHz linear arrays. A pulser is dedicated
to each element so that only low voltage trigger signals are necessary to drive
16 to 48 demultiplexers. A microprocessor (89S8252) based control board is
used to generate all the necessary control signals for multiplexers and
demultiplexers in the tranceiver board and analog delaylines in the beamforming
board. A 40-60 dB TGC amplifier is applied to echoes from each channel before
they are aligned by a cross-point switch and analog delaylines. One channel
beamformed echo signals are digitized by a 500MS/s 8-bit PCI A/D card in PC.
Finally the acquired data are processed and B-mode images are displayed in

79
real time by Labview based software. The system can display 30 images per
second and acquire over 100 images per second for slow motion playback by
using a 48 element linear array with center frequency at 30MHz. By replacing
with more powerful microcontrollers and more efficient C++ based software, it
should be possible to acquire images at a much higher frame rate.
Rabbit eyeball images and mouse heart images have been obtained in vitro
and in vivo to demonstrate the potential of this system in biomedical applications.

P1E-7
ULTRASOUND-BASED AIR BUBBLE TRAPPING
SYSTEM FOR HAEMODIALYSIS.
P. PALANCHON*1, A. BOUAKAZ2, L. POURCELOT3, and F. TRANQUART2, 1CIT,
CHU Bretonneau, Tours, France, 2INSERM U619, Tours, France, 3Ultrasound
and Nuclear Medecine, CHU Bretonneau, Tours, France.
Corresponding e-mail: palanchon@[Link]
Background: Microembolization in patients undergoing chronic haemodialysis
sessions is considered as a potential source of severe pulmonary side-effects.
Previous experimental and clinical studies indicated that several air bubbles
were generated by the haemodialysis system but they were not always trapped
by the air filter. The goal of this study was to develop an air bubble trapper using
ultrasound waves to remove any bubble from the tubing system before they
reach the patient.
Method: A home-made bubble trapper was developed in the laboratory. It consists
of a Perspex block containing a main channel connected to the tubing of a
haemodialysis machine and a second sub channel perpendicularly positioned
to the main one. The sub-channel is used to trap the air bubbles. The bubbles
flowing in the main channel were insonifed through an acoustic window with an
ultrasound wave, pulsed (8 to 20 cycles) or continuous, at a frequency of 500
kHz generated by a single element transducer positioned 3cm away from the
main flow. The generated acoustic pressure was kept below 1MPa. The radiation
force induced by the ultrasound beam acts directly on the flowing air emboli, by
pushing them into the sub-channel. Two Doppler probes operating both at 2.5
MHz, connected to a DWL Doppler machine were placed before and after the
bubble trapper to count simultaneously the number of embolic events. The flow
of the machine was varied between 200ml/min and 500ml/min.
Results : Depending on the flow velocity, the number of microembolic signals
(MES) detected by the Doppler probes before and after the trapping system
was identical and ranged 5 to 150 MES/min in absence of the ultrasound
irradiation. When the air bubble trapper was activated, a reduction of the number
of MES detected by the second Doppler probe of up to 70% was observed.
According to the Doppler recordings, the circulating bubbles were either
fragmented into smaller bubble fragments or directly got pushed into the second
sub-channel where they were collected.
Conclusions: This simple approach using an ultrasound-based trapping system
showed to operate adequately with the current settings and can be used to
neutralize air emboli.

80
P1E-8
VERSATILE ASSESSMENT OF 3D PROSTATE
ANATOMY AND ELASTIC ANOMALIES.
Y LI*, C. D. GARSON, and J. A. HOSSACK, University of Virginia, Charlottesville,
VA.
Corresponding e-mail: yinbo@[Link]
Prostate cancer is the second most prevalent malignant cancer among men in
the US with approximated 220,000 new cases and 29,000 deaths per year [1].
Unfortunately, current prostate cancer screening approaches using digital rectal
examination and Prostate Specific Antigen (PSA) blood testing, yield
disappointing diagnostic sensitivity and specificity. In this work, we use a modified
transrectal transducer operating in the ‘I-Beam’ mode [2]. Auxiliary tracking arrays
provide image frame to frame measurement of offset and rotation facilitating
later reconstruction into a 3D image data set. This particular approach to 3D
positioning is especially well suited to this application since the 3D positioning
device is adjacent to the B-scan imaging transducer at the tip of the transducer
probe. This avoids the numeric ill-conditioning that may be encountered when
the position measurement device and image plane are physically separated as
they would be if the position measurement device is in the handle of a transrectal
transducer. Additionally, the I-Beam approach senses relative tissue motion and
thus any bulk tissue motion during the scan is at least partially compensated
out in the final reconstruction. The approach is also compatible with the generally
preferred freehand scanning technique (as opposed to approaches based on
using a mechanized stage). In 3D reconstructions of lesions in a realistic prostate
phantom we have achieved linear dimensional accuracy of 5% in the
reconstructed (elevational) dimension and lesion volume measurement accuracy
of 11%. The modified transducer is also fitted with a water injection port to facilitate
a technique that we refer to as ‘Synthetic Digital Rectal Examination’. (This
approach was independently conceived by Alam et al. and described at the
2004 IEEE Ultrasonics Symposium. [3]) The latex sheath over the transducer is
inflated using a syringe pump. In prototype phantom testing, we have verified
our ability to detect 3 × 3 × 5mm lesions via the computed elastographic image
at a depth of 2 cm. Thus, we believe that we are able to detect and quantify
lesions that are small smaller and deeper than those that may be practically
palpated using conventional digital rectal examination.
[1] American Cancer Society, “Prostate Cancer Statistics from [Link],”
2003.
[2] J. A. Hossack, T. Sumanaweera, S. Napel, and J. Ha, “Quantitative 3D
Diagnostic Ultrasound Imaging Using a Modified Transducer Array and an
Automated Image Tracking Technique,” IEEE Transactions on Ultrasonics
Ferroelectrics & Frequency Control, vol. 49, pp. 1029-1038, 2002
[3] S. Alam, E. Fellepa, A. Kalisz, S. Ramchandran, R. Ennis, F. Lizzi, C.-S.
Wuu, and J. Ketterling, “In vivo Prostate Elastography: Preliminary Results,”
Proceedings of IEEE Ultrasonics Symposium, 2004.
This work supported in part by US Army Grant W81XWH-04-1-0240.

81
P1E-9
A STUDY OF MOTION ARTIFACTS OF FOURIER-
BASED IMAGE CONSTRUCTION.
J. WANG* and J.-Y. LU, Ultrasound Lab, Dept. of Bioengineering, The University
of Toledo, Toledo, OH.
Corresponding e-mail: jilu@[Link]
Based on high frame rate method developed in our lab, a Fourier based imaging
method with a variable frame rate was developed recently. In this method, multiple
steered plane waves are used to obtain ultrasound echo signals. Images are
constructed based on Fourier transformation. Because multiple transmissions
may be used to obtain a frame of image to reduce sidelobe and increase
resolution, it is important to study the effects of motion on the method for fast
moving objects such as the mitral valve of the heart, and compare the results
with those of conventional delay-and-sum method.
In this paper, we performed experiments with the Fourier based method by using
our high frame rate imaging system to obtain radio frequency (RF) data. A point
scatterer was placed in the imaging plane, moving perpendicularly to the axis of
a 2.5MHz, 19.2mm aperture, and 128 element array transducer at a velocity of
210 mm/s which is about the peak velocity of the mitral valve of a heart.
Experiments were repeated with the point scatterer placed on axis but with
different depths (30, 46, 50, 70 and 90mm) from the transducer surface. At each
depth, data from different numbers of transmissions (1, 11, 19, and 91) were
used to construct images with the Fourier-based method (frame rates of 5346,
486, 281, and 59 frames/s, respectively). As a comparison, images were also
constructed with conventional delay-and-sum method with and without dynamic
focusing in transmission. (For delay-and-sum, the frame rate is about 59 frames/
s with a fixed transmission focal distance at 70 mm. For dynamic focusing at all
depths in transmission, the frame rate would be very low but the image quality
would be high.) To study the image contrast changes, an ATS539 tissue mimicking
phantom was in place of the point scatterer with the center of its six 15mm
diameter cylinders located at a depth of 46mm. Images constructed without
object motion were also obtained and were used as bases to calculate
corresponding percentage changes of resolution, sidelobe, and contrast.
The results show that for 1, 11, and 19 transmissions, the percentage changes
for the Fourier-based method are smaller than those of delay-and-sum method
without dynamic focusing. For 91 transmissions, the percentage changes are
larger in small depths but are comparable with the delay-and-sum method when
distance increases. In conclusion, the Fourier-based method is not very sensitive
to the motion except when the number of transmissions is large (lower frame
rate) and the depth is small.
The authors would like to thank Mr. Jiqi Cheng for his help in doing the
experiments and providing the image construction programs. This work was
supported in part by a grant HL 60301 from the National Institute of Health.

82
Session: P1F
ACOUSTIC SENSORS
Chair: P. Khuri-Yakub
Stanford University
P1F-1
INVESTIGATION OF PRECISION SOUND VELOCITY
MEASUREMENT METHODS AS REFERENCE FOR
ULTRASONIC GAS FLOW METERS.
P. NORLI*1, P. LUNDE1, and M. VESTRHEIM2, 1Christian Michelsen Research
AS (CMR), Bergen, Norway, 2University of Bergen, Dept. of Physics and
Technology, Bergen, Norway.
Corresponding e-mail: pnorli@[Link]
Ultrasonic gas flow meters for volumetric flow rate fiscal metering of natural gas
(USMs) may today also be used for mass and energy flow rate measurement,
based on velocity of sound (VOS) measurement. To establish the accuracy of
the VOS measurements given by the USM, and for traceability purposes, an
independent and high-accuracy VOS measurement cell may be used as
reference. To include relevant effects of dispersion, the cell should preferably
work in the operational frequency range of USMs, 100-200 kHz, with natural
gas under high pressure. VOS measurement cells with extreme accuracy are
available in the audio frequency range, for which uncertainties down to 1 ppm
have been reported. Less work is however identified at a sufficient accuracy
level in the frequency range 100-200 kHz.
Three different candidate transient methods have been tested and analyzed
with respect to measurement uncertainty, and they are seen to have several
common experimental uncertainty sources. In the present work, a two-distance
method is discussed in more detail as an example, and some results from
measurements in an insulated chamber with air at 1 atm and ca. 25 °C are
presented.
The relative expanded measurement uncertainty was estimated according to
ISO guidelines to approximately 264 ppm (95% conf. level). One major source
of uncertainty in these measurements was experienced to be small convection
currents in the chamber. Without these, the expanded measurement uncertainty
would have been about 126 ppm. Such convection effects are however expected
to be strongly reduced in a properly designed measurement cell.
The VOS measurement results were compared with predictions from Cramer’s
VOS model for standard air, including dispersion [J. Acoust. Soc. Am. 93 (5),
pp. 2510-2516, 1993], resulting in a mean deviation of -18 ppm with a two
standard deviation spread in the data of 190 ppm.

83
P1F-2
A CONTROLLED INVESTIGATION OF THE SPATIAL
DISTRIBUTION OF ACOUSTIC CAVITATION ACTIVITY
GENERATED IN A STANDING WAVE FIELD
PRODUCED BY A 40 KHZ CLEANING
VESSEL TRANSDUCER.
M. HODNETT*, M. CHOI, P. GELAT, and B. ZEQIRI, Quality of Life Division,
National Physical Laboratory, Teddington, Middlesex, UK.
Corresponding e-mail: [Link]@[Link]
High power ultrasound has found significant utility in a wide range of applications,
from healthcare, through sonochemistry, to cleaning and industrial processing.
The phenomenon of cavitation is the principal driver behind many of these
applications, yet quantification techniques for cavitation are still not widely
available. Access to such measurement methods, and their establishment into
a standards infrastructure would undoubtedly lead to process improvements
and enhanced technology take-up.
The work reported describes work carried out at NPL to provide a facility against
which the performance of NPL’s patented novel cavitation sensors may be tested
on a quantitative, reproducible basis. It therefore comprises the results of a
systematic measurement programme carried out on a 40 kHz single transducer
system, built around a stressed-stack piezoelectric transducer typical of the
type used in ultrasonic cleaning systems. Both a conventional sonar hydrophone
(B & K type 8103) and NPL cavitation sensors have been used to measure the
distribution of acoustic pressure and cavitation activity generated by the 40 kHz
transducer in a cylindrical vessel. Repeat measurements show that the acoustic
pressure amplitude, even under cavitating conditions, is reproducible to ±15%.
Cavitation activity, defined as the broadband acoustic emissions from bubble
oscillation and collapse integrated over the frequency range 1 - 3 MHz, is
reproducible to ±20%. Both cavitation activity and acoustic pressure
measurements have been carried out for a range of water column heights and
transducer drive voltages, and the results illustrate clearly the evolution of
cavitation activity with increasing applied acoustic pressure, and its enhancement
when standing waves are established.
To predict the acoustic pressure distribution in the cylindrical vessel, a detailed
Finite-Element (FE) model was set up of the transducer and vessel combination.
Results from this show good agreement with experiment in terms of the locations
of acoustic field maxima and minima, and the model has thus been used to
refine the experimental protocol adopted when testing sensors experimentally.
Finally, the motion of the emitting face of the 40 kHz transducer in water has
been probed experimentally using a Polytec scanning laser vibrometer. The
results obtained show that the vibration characteristics of the 40 kHz transducer
are approximately piston-like, but that future FE models will need to consider a
more detailed approach to transducer construction, in particular allowing for the
effects of the tensioning bolt.
The authors acknowledge the financial support of the National Measurement
System Directorate of the UK Department of Trade and Industry.
84
P1F-3
INVESTIGATION ON VISCOELASTICITY OF SILICONE
RUBBER USING IMPEDANCE CHANGE OF A QUARTZ-
CRYSTAL TUNING-FORK TACTILE SENSOR.
H. ITOH* and Y. YAMADA, Shinshu University, Nagano, Japan.
Corresponding e-mail: rokuro1@[Link]
It is well known that viscoelastic data of silicone rubber measured by rotating
viscometer or DMA (Dynamic Mechanical Analysis) is quite different each other.
Viscoelastic data strongly depends on measurement equipment. In this paper,
viscoelasticity of silicone rubber has been investigated experimentally by use of
a quartz-crystal tuning-fork tactile sensor. The quartz-crystal tuning-fork tactile
sensor makes use of its impedance or frequency change at resonant vibration
32.48 kHz when its base gets brought into contact with an object. The quartz-
crystal tactile sensor has many advantages. It is capable of distinguishing a
wide variety of materials from soft to hard ones by use of the impedance change
of the quartz-crystal tactile sensor. Their quality and surface roughness are also
distinguished. The experiments were done for eight kinds of silicone rubbers
(the values of rubber hardness tester are JIS85, 70, 65, 60, 50, 45, 40, and 35).
Rubber hardens as the value of rubber hardness tester increases. The impedance
change (∆ R) was calculated between the sensor in contact with an object and
in no contact at room temperature. The impedance change increases according
to the value of rubber hardness tester. At the same time, the impedance change
increases according to acoustic impedance ρ C (ρ: density of an object, C:
sound velocity of longitudinal acoustic wave in silicone rubber) because the
impedance change has been proportional to the transmitting energy resulting
from the difference between silicone rubber and the base of the quartz-crystal
tuning fork in acoustic impedance of the longitudinal plane wave. We compare
two cases of impedance change characteristics: one is obtained by C calculated
by Young’s modulus measured by tensile meter, the other is calculated by
complex Young’s modulus measured by rotating viscometer. In the range of low
ρ C of silicone rubber compared to metals, the impedance change is almost
proportional to ρ C. The correlation coefficient of linearity between ∆ R and ρ C
for the former case is 0.89 and for the latter case is 0.929. It is found that the
impedance change of a quartz-crystal tuning-fork tactile sensor correlates closely
with complex Young’s modulus of silicone rubber because silicone rubber is
intrinsic viscoelasticity and includes the information of viscoelasticity of silicone
rubber as its complex Young’s modulus.

P1F-4
INFLUENCE OF RECEIVER NOISE PROPERTIES ON
RESOLUTION OF PASSIVE WIRELESS RESONANT
SAW SENSORS.
V. KALININ*, Transense Technologies plc, Upper Heyford, Bicester, Oxon., UK.
Corresponding e-mail: [Link]@[Link]
Wireless or contactless sensors based on SAW resonators and delay lines
working as passive back-scatterers have found application in the areas where
85
temperature and mechanical strain need to be measured on rotating parts. As
an example, resonant SAW sensors are used in automotive tire pressure
monitoring systems (TPMS), they are being developed for torque measurements
in electrical power assisted steering systems, driveline control and engine
management systems.
Resolution is one of the most important characteristics of the passive wireless
SAW sensor. It is mainly determined by noise in the measured value, which in
its turn depends on the loaded Q factor of the SAW resonator, noise properties
of the receiver and the algorithm used for spectrum estimation. Several
publications investigated theoretical limit for the resolution that is set by additive
noise of the receiver. It was shown that the potentially achievable resolution
could be very high. For instance, the standard deviation of the measured strain
can be as small as 0.02 microstrain at a distance of 1 m for the interrogation
power of 10 mW at 433 MHz, the receiver bandwidth of 0.5 MHz, noise figure of
5 dB and the loaded Q = 6000 of the resonator made on ST-X cut quartz. However
in reality the achievable resolution is considerably worse. This paper investigates
additional limitations on the SAW sensor resolution that are imposed by the
phase noise of the local oscillator of the receiver.
The problem is studied by means of stochastic simulations on the basis of
information on the realistic phase noise of the receiver synthesiser used in the
wireless interrogation unit. It is shown that the additive noise determines the
resolution only at a distance larger than 3 m in the free space. This situation
may be typical for TPMS where the received signal power can be as low as -80
dBm. The received signal in torque sensors is much stronger, typically from -20
to -10 dBm. In this case or in the case of shorter distances for TPMS it is mainly
the phase noise that determines the resolution. For a typical value of the phase
noise of -90 dBc/Hz at 10 kHz offset the predicted standard deviation of measured
strain is 25 times larger than the one determined only by additive noise. This
figure is much closer to experimental results.
The paper also studies variation of the dynamic range of the wireless resonant
SAW sensor with interrogation distance and the number of coherently
accumulated SAW responses. Practically achievable dynamic range for the
torque sensor is close to 60 dB.

P1F-5
PIEZOELECTRIC BIMORPH MICROCANTILEVER: A
NEW GAS PRESSURE SENSOR.
V. MORTET*1, R. PETERSEN2, K. HAENEN1,2, and M. D’OLIESLAEGER1,2,
1
Universiteit Hasselt, Institute for Materials Research, Diepenbeek, Belgium,
2
IMEC vzw, Division IMOMEC, Diepenbeek, Belgium.
Corresponding e-mail: vmortet@[Link]
Since the development of the atomic force microscope (AFM), the interest in
micro-fabricated cantilevers has grown. Micro-machined cantilevers are
extremely sensitive, miniature, mass produced and low cost sensors. Cantilevers
are excellent micro-mechanical sensors. Micro-cantilevers sensors operate by
detecting changes either in resonance frequency, amplitude, Q-factor or

86
deflection caused by mass loading, surface stress variation, or changes in
damping conditions. In this work, we used a non-symmetric piezoelectric bimorph
cantilever as a wide range pressure sensor. The sensor detects the change in
the resonance frequency of the micro cantilever with a piezoelectric film (ZnO).
In contrary to common cantilever sensor systems, which use either optical,
integrated piezo-resistive, integrated piezoelectric, or integrated capacitance
detections, the piezoelectric film of the bimorph cantilever acts as both a sensor
and an actuator. The frequency characterization of the piezoelectric bimorph
cantilever was accomplished by measuring the bimorph’s admittance using an
impedance/gain-phase analyzer.
In this communication, we report the experimentally measured changes in the
cantilever’s resonant frequencies versus the gas pressure (up to 7 bar), the
type of the gas and the temperature (up to 60°C) and we discuss the optimisation
of both mechanical and geometrical properties of the cantilever for pressure
measurement.
The cantilever exhibits several resonance frequencies. We have measured
variation of the two first resonance modes at f1 =46kHz and f2 =913.4kHz (at
room temperature and atmospheric pressure). Both resonance frequencies show
a linear variation as a function of the pressure and they exhibit the same variation
as a function of the temperature. The frequency shift as function of the pressure
is -0.00421 /bar for the first mode and -0.00319 /bar for the second mode. These
experimental results show that a simple cantilever can be use to measure both
pressure and temperature simultaneously.

P1F-6
A NOVEL DOPPLER BASED ULTRASONIC SURFACE
ROUGHNESS MEASUREMENT.
J. REZANEJAD GATABI*1 and I. REZANEJAD GATABI2, 1Iran University of
Science and Technology, Tehran-Tehran-Iran, 2Khajeh Nasir Toosi-University of
Technology, Tehran-Tehran-Iran.
Corresponding e-mail: irezanejad@[Link]
Surface topography is of great importance in specifying the function of a surface.
A significant proportion of component failure starts at the surface due to either
an isolated manufacturing discontinuity or gradual deterioration of the surface
quality. Typical, these problems of lower surface integrity, lead to in service
stress corrosion and fatigue failure. The most important parameter describing
surface integrity is surface roughness.
In the manufacturing industry, surface must be within certain limits of roughness.
Therefore, measuring surface roughness is vital to quality control of machining
workpiece, especially if it is non-contact method compared to the direct
conventional method; which uses stylus type devices.
In this paper, a novel Doppler based ultrasonic sensor has been proposed for
high-precision surface roughness measurement. Despite conventional non-
contact ultrasonic techniques which use the ultrasonic travel time to measure
the distance of the undertest surface from the sensor to describe the roughness
parameters, the proposed sensor implements the Doppler effect on the

87
continuously emitted ultrasonic bursts for surface roughness measurement.
When an object is ready to undergo roughness testing, it is mounted on a motor
rotating at a constant velocity of ω. The device comprises an ultrasonic transmitter
emitting sound pulses travel across to the undertest object. The sonic wave is
reflected diffusely on the object where it is separated into many weak sounds, a
few of which are received by the receiver. The Doppler effect causes the
frequency of the received signal to be shifted with respect to the roughing depth
and its shape. The relationship between the rate of change of the frequency of
the received signal and the peak-to-valley height and the roughing shape is
mathematically analyzed.
After amplification, the received signal is applied to a high-pass filter and therefore
to a rectifier stage. Of its frequency response, the circuit provides an output
voltage related to the frequency of the received signal. Analysis of the output
voltage by a processor allows the parameters of the surface roughness to be
calculated.
Since the sonic speed changes in various environmental situations, it may cause
some errors in transit-time based sensors. Using Doppler effect, these problems
are avoided and therefore a more precise measurement is achieved. Also the
continuous ultrasonic emission allows the processor to sample the surface at
any point where needed.
The proposed method also can be applied for vibration analysis of the rotating
fans. The fabrication of a prototype sensor and experimental verification of the
analytical results are reported. The experimental results agree well with the
theory.
With specially thanks to following people for their assistance and encouragement:
1- Dr. F. Forouzbakhsh- University of Tehran 2- Mr. A. Araghi- University of
Mazandaran

P1F-7
INTERACTION OF SURFACE ACOUSTIC
WAVES WITH PARTICLES.
J. KONDOH*1, T. OYAMA1, Y. MATSUI1, and S. SHIOKAWA2, 1Faculty of
Engineering, Shizuoka University, Hamamatsu-shi, Shizuoka , Japan,
2
SAW&SPR-Tech, Hamamatsu-shi, Shizuoka, Japan.
Corresponding e-mail: j-kondoh@[Link]
This paper presents interaction of surface acoustic waves (SAW) and particles.
Liquid-phase sensor is realized by using a shear horizontal SAW (SH-SAW).
Measurements of particles in liquid are important application for biosensor,
process monitoring, and so on. The detection mechanisms of the SAW are
mechanical and electrical perturbations. For simultaneous detection of both
perturbations, three-channel SH-SAW sensor was proposed. Several particles,
such as pigments, guard gel (diameter is 10 micron), and steel ball, are measured.
The results indicate that it is difficult to detect particles in liquid based on the
mechanical perturbation. We have concluded that the penetration depth of the
shear horizontal displacement is not enough to detect particles, because it is
about 80 nm, when water is loaded on 50 MHz SH-SAW sensor. As the viscous

88
penetration depth increases with decreasing sensor frequency and increasing
viscosity, the steel balls are immersed in 80 wt.% glycerol/water mixture. The
results show that the high sensitive detection is realized with 30 MHz SH-SAW
sensor.
The measurements of mass loading effect in gas phase are performed by using
the steel balls. The steel balls, whose weight is a few grams, are loaded on the
SH-SAW propagating surface. Whereas the mass loading effects in liquid are
not observed, phase, i.e. frequency, increment phenomenon is observed in the
gas phase. On the other hand, amplitude of the SH-SAW is not changed.
Increment of phase corresponds to increase of the phase velocity. The same
phenomenon is observed by using Rayleigh-SAW sensor. Based on these
results, we have assumed that the elastic constants of the piezoelectric crystal
increase by loading of the steel ball. This assumption agrees of a pressure
sensor based on the acoustic wave sensor. The results suggest that a threshold
exist between the mass loading effect and pressure effect.
This work is supported by Human Frontier Science Program.

P1F-8
ORGANIC VAPOR SENSING AND DISCRIMINATION
USING ENHANCED SENSITIVITY THICKNESS SHEAR
MODE DEVICES.
R. WILLIAMS, S. CULAR*, A. UPADHYAYULA, and V. BHETHANABOTLA,
University of South Florida, Tampa, FL.
Corresponding e-mail: willia15@[Link]
Thickness shear mode (TSM) resonators, also known as quartz crystal micro-
balances (QCM) are a class of acoustic wave sensors that have been used for
gas/vapor sensing and for determining liquid properties. Fast and sensitive
chemical vapor sensing, specifically of hydrocarbon vapors, has attracted recent
attention. The TSM sensors typically used have a lower sensitivity compared
with other acoustic wave sensors. This paper describes the development of
high sensitivity organic vapor sensors using polymer thin film coatings of poly-
isobutylene (PIB) on TSM devices. Commercially available AT-quartz TSM
devices were milled to 17 µm, leaving a thin quartz membrane surrounded by a
50 µm thick outer ring. This resulted in an increased frequency and a consequent
increase in sensitivity, as described by device models. TSM devices with
fundamental mode resonant frequencies of 10, 20 MHz were compared to the
milled 96 MHz devices. The organic vapors studied were benzene, toluene,
hexane, cyclohexane, heptane, dichloroethane, dichloromethane, and chloroform
at levels ranging from less than 1 to over 10 volume percentage in nitrogen gas.
The Butterworth-VanDyke (BVD) equivalent circuit model was used to model
both the perturbed and unperturbed TSM resonator. Monitoring the sensor
response through the equivalent circuit model allowed for discriminating between
the organic vapors. In particular, changes in the resistance parameter due to
softening and relaxation of the PIB film allowed for this vapor discrimination. For
instance, it was found that vapor discrimination of benzene vapors was possible
at over 4.74 % levels which corresponded to resistance changes over 5 ohms.

89
We present results of tests conducted to demonstrate increase in sensitivity for
higher fundamental frequency TSM devices. We also evaluate and compare
the performance of each sensor in terms of detection limit and noise level.
Inverse mesa resonators were fabricated at MTronPTI, Orlando Florida. The
authors acknowledge the significant contributions of Shibendra Pobi, Krishnan
Srinivasan, and Tom Payne.

Session: P1G
NDE MODELING AND MEASUREMENTS
Chair: N. Bilgutay
Drexel University
P1G-1
STUDIES ON EFFECTIVE AND STABLE ABSORBING
BOUNDARY CONDITIONS IN ULTRASONIC
WAVE MODELING.
H. ZHAO1, X. WANG1,2, and H. ZHANG*1, 1Institute of Acoustics, Haidian District,
Beijing, China, 2CSIRO Petroleum, Bently, WA, Australia.
Corresponding e-mail: [Link]@[Link]
In ultrasonic wave numerical modeling for inhomogeneous material, it is
impossible to allow numerical simulation of ultrasonic wave propagation in
unbounded medium with a limited computation model. Usually the ultrasonic
waves in an unbounded medium are simulated using bounded models through
imposing artificial boundaries. These artificial boundaries, if not treated properly,
may cause spurious unwanted reflections that affect the accuracy of numerical
solutions. Thus, effective and stable absorbing boundary conditions (ABCs) to
reduce or eliminate artificial reflections from these artificial boundaries are
needed. The perfectly matched layer (PML) has been applied to both acoustic
and elastic problem for simulating wave propagation. It is known that apparent
reflections occur at the interface between inner normal regions and PML regions
are reduced greatly.
In this paper, in order to conduct effective ultrasonic wave numerical experiment
in inhomogeneous material for perfect artificial reflection absorptions, a high-
order staggered-grid finite-difference scheme, with velocity-strain equation
systems is proposed. This algorithm uses the PML method, the damping factor
method, and the Liao’s method that are considered to be more efficient than the
others. These three ABCs are carefully implemented in a staggered-grid finite-
difference, and studied in detail. Numerical examples for the three kinds of
absorption boundary conditions are run, and the related numerical results are
compared with each other. The related algorithms are implemented by means
of calculating reflection coefficient in the case of the discrete condition. The
results both in time and in space domains show that the PML ABC is better than
either of damping factor ABC and Liao’s ABC for all of the reflected angles.
When the Lao’s and damping factor methods are used, the reflections become
much stronger as reflected angle is larger than 50 degree. In addition, we also

90
draw a conclusion that the PML is highly efficient in absorbing the surface waves
particularly.

P1G-2
MEASUREMENT METHOD OF LONGITUDINAL
ACOUSTIC PROPERTIES FOR SOLID SPECIMENS
USING THIN LAYER WATER COUPLANT
IN UHF RANGE.
H. ODAGAWA*, M. ARAKAWA, K. MORIOKA, and J. KUSHIBIKI, Tohoku
University, Sendai, Japan.
Corresponding e-mail: odagawa@[Link]
High frequency and wide band ultrasonic measurement methods are required
in ultrasonic material characterization to evaluate velocity dispersion and
frequency characteristics of attenuation coefficient. Generally, measurements
for solid specimen are performed in the ultrasonic composite transmission line,
consisting of a buffer rod with an ultrasonic transducer on one end, a couplant
and a specimen. Pure water is usually used as the couplant, because precise
acoustic properties was obtained: no velocity dispersion at least 1GHz, known
attenuation coefficient (α/f 2 =22.3×10-15 s2/m) and velocity (V=1491.23 m/s) at
23°C. However, because of large attenuation of an ultrasonic wave in the water,
it is difficult to use it in UHF range by the conventional method. A method using
salol (phenyl salicylate) as a solid couplant has been reported for UHF range
measurement. However, this method has a complexity that the acoustic
characteristics of the salol couplant depend on the solidification condition.
Therefore if the water couplant can be applied in UHF range, it is very useful to
characterize the acoustic properties of materials.
In this paper, we will describe a measurement method of UHF-range longitudinal
acoustic properties for solids using a thin water layer as a couplant. The thickness
of the couplant t is reduced to around 1µm to reduce the propagation attenuation.
It is approximately half wavelength of the center frequency of the measurement
range. Consequently, multiple reflections of the acoustic waves are occurred in
the couplant. Reflection coefficient at the water couplant from the buffer rod
becomes minimum in the frequency fd satisfying a condition of fd =V/(2t). We
can determine the accurate value of t by measuring fd. Considering them, we
develop an analysis method using a fitting technique. In this measurement, a
RF pulse signal and complex mode are used.
In order to confirm a validity of this method, some measurements have been
demonstrated for synthetic silica glass which has no velocity dispersion in VHF
and UHF ranges; T-4040 (Toshiba Ceramics Co.) and N-ES (Nippon Silica Glass
Co.) were measured 300-750 MHz and 450-1100 MHz, respectively. We made
several ultrasonic devices, which had piezoelectric ZnO thin film transducers
on the one side of the buffer rods and spacers made by ZnO thin film on the
opposite side. The spacers kept the thickness of the water couplant constant as
well as the parallelism of the thin layer. The devices and specimens were put in
a small chamber, which soaked in a temperature controlled water bath. The
results showed a good agreement with the value measured by our conventional

91
methods in VHF and UHF ranges. Velocities of T-4040 and N-ES were 5953.4±0.2
m/s and 5927.2±0.1 m/s, respectively. The deviations from the previous results
was less than ±0.004%, and they exhibit no velocity dispersion up to 1GHz.
Measured α/f2 of these specimen were 1.2-1.4×10-16 s2/m (previous results:
1.3×10-16 s2/m). Therefore, we confirmed that this method was very useful and
powerful for characterizing solid materials.

P1G-3
APPLICATION OF THE FDTD TF/SF METHOD TO
ANALYSIS OF THE SCATTERING PHENOMENA OF
THE ELASTIC WAVE FIELDS IN SOLIDS.
M. SATO*, Akita University, Akita, Akita, Japan.
Corresponding e-mail: ma-sato@[Link]
Scattering phenomena of the elastic wave fields in solids is the most critical
aspect to be clarified in the non-destructive evaluation (NDE) of solids using
ultrasonic waves. In order to obtain the desired information from the echo signals
of elastic waves reflected from the flaws in solid materials, the relation between
the signals and the flaws should be clarified. For this objective, analysis by
numerical simulation represents the most versatile tool available because of
the difficulties in investigating by other analytical and experimental methods. By
using numerical methods such as the finite difference method (FDM), finite
element method (FEM) and boundary element method (BEM), researchers have
clarified various characteristics of the scattered elastic waves.
The present author and colleagues have previously reported the application of
a Finite-Difference Time-Domain (FDTD) numerical method to the elastic wave
fields in solids. The FDTD method has the unique characteristic that it can
separately calculate the far field scattered waves and near field total waves at
the same time, where total waves refer to a mixture of incident plane waves and
scattering waves. This formulation is called the total-field/scattered-field (TF/
SF) formulation and is very effective for the analysis of scattering problems in
NDE.
Generally, perfect plane waves are difficult to realize in numerical calculations.
If we input the plane-like wave in the analysis region, both of the sides of the
wave will be diffracted. Also, if we realize the plane wave by adopting some
boundary conditions at the truncated boundary, we will not be able to absorb the
scattering waves at the boundary. The TF/SF formulation, however, can realize
perfect plane waves that are incident to the scattering objects.
To date, the TF/SF formulation has only been applied to scalar wave fields and
electromagnetic wave fields. Application to the elastic wave fields in solids has
not yet been attempted.
In this report, a TF/SF formulation of the elastic wave fields in solids is described.
Specifically, scalar and vector potentials are used for the formulation. Using
these potentials is beneficial for two reasons. First, the longitudinal waves and
shear waves are initially separated. Therefore the scattering phenomenon can
be clearly recognized. Second, it is facile to use the potentials for setting the
absorbing boundaries. Generally, it is not simple to implement the absorbing

92
boundaries of the elastic wave fields in solids. Because, the two waves,
longitudinal and shear waves, exist in the solid material and the waves have
different characteristics, it is difficult to match the absorbing boundaries to both
waves at the same time. However, if we use the potentials, we can divide the
waves perfectly and set the boundary to each wave separately such as in a
scalar wave.
The FDTD TF/SF formulation has been proved effective in analysis of the
scattering phenomenon of a longitudinal plane wave incident to a square hole.

P1G-4
STUDY OF THE TIME-DEPENDENCE OF THE
MECHANICAL PROPERTIES OF DOUGHS FOR FLOUR
STRENGTH EVALUATION.
J. GARCIA-ALVAREZ*, J. M. RODRIGUEZ, Y. YAÑEZ, A. TURO, J. A. CHAVEZ,
M. J. GARCIA, and J. SALAZAR, Technical University of Catalonia (UPC),
Barcelona, Spain.
Corresponding e-mail: [Link]-alvarez@[Link]
Both flour strength and dough processing affect the dough consistency that
determines its potential for breadmaking purposes. Quick identification of poor
dough quality would reduce problems with dough handling during further stages
of the process and maximise productivity. Maintaining consistent production
would contribute to better control of product quality and consequently lead to
high levels of customer satisfaction.
Ultrasonic measurements have already been carried out to characterise dough
properties. The ultrasonic wave parameters generally measured include the
velocity of propagation and the attenuation of the acoustic wave travelling through
the sample. These measurements can be related to both viscoelastic and physical
properties of the sample, providing the flour strength. However, due to the time-
dependent nature of dough accurate measurements of the ultrasonic velocity
and attenuation are sometimes difficult to attain especially in highly attenuating
materials like dough. Furthermore, due to viscoelastic properties of dough when
a sample is placed between both transducers it slowly flows away from the
transducer surface producing changes in the values of the ultrasonic velocity
and attenuation with time. The greater these changes are the softer the dough
is. This makes necessary a settling time in order to get an accurate measurement.
In this work, an alternative method for evaluating the flour strength using low
intensity ultrasound is shown. The evolution with time of both velocity and
attenuation is monitored and then related to the flour strength. Main advantage
of this novel approach is that changes in time of ultrasonic velocity and attenuation
are easy to monitor than carry out accurate measurements of them after a settling
time. Experimental results on doughs with different flour strength are presented,
compared and discussed.
This work was supported by a CRAFT Project under contract no. QLK1-CT-
2001-70377. Also, the authors would like to thank the baking and Cereal
Processing section of CCFRA for dough preparation, bread baking and bread
assessment.

93
P1G-5
DESIGNING AND SIGNAL PROCESSING OF
INTELLIGENT INSPECTION PIG APPLYING
ULTRASONIC A-SCAN.
C. TIANLU*, Q. PEIWEN, J. TAO, and Z. ZHIGANG, Dept. of Information
Measurement technology and Instrument, Shanghai Jiaotong University,
Shanghai, P. R. China.
Corresponding e-mail: chentianlu@[Link]
Pipeline safety evaluation is an issue of great concern in China. Several types
of NDE methods have been tried and tested for inspecting pipelines. Great
success has been achieved using ultrasonic A-scan to size and locate defects
in oil and gas pipelines. An ultrasonic inspection system (intelligent pig) is
designed to inspect pipelines of different diameters. The overall architecture,
hardware configuration, and signal processing strategy are discussed.
The intelligent pig (2400mm long, 60 kg weight) includes six parts: driver robot,
system controller, power supply, ultrasonic sensors, ultrasonic data processor,
and position tracer. The robot can drive the pig to climb in pipeline from 150 mm
to 350 mm diameter with speed about 15-20 mm/s. 128 ultrasonic transducers,
a humidity sensor, a temperature sensor and a pressure sensor are fixed on
special sensor carriers and arranged scientifically to ensure that the stand-off
and the transmission angle of the sensors are fixed and the whole pipeline wall
can be covered. The CAN bus communication network inside the pig is
represented by a block diagram. Multi-channel parallel inspection procedure
and the sensor trigger sequence are introduced. Inspected signals are digitized,
selected, compressed by data processors and stored in four hard disks for off-
line analysis. The position data generated by the odometer is stored together
with corresponding A-scan signal.
In lab, we use the pig to inspect samples which have same material and size
with Chinese offshore pipelines. Defects as flaw, crack and corrosion with different
size made by electricity-sparkle are prepared. The inspection signal analysis
procedure is described in detail step by step. Downloaded and decompressed
raw signals are firstly de-noised and compensated by wavelet theory. Then,
defect characters such as size, location and type can be extracted through
temporal and spatial alignment, association, and wavelet neural network fusion.
They are stored in defect character database. Full three-dimensional profiles of
defects are given. Finally, fuzzy ARTMAP is used to classify defects into five risk
levels according to ASME B31G. The classifier is trained by pipeline defects
information provided by Shengli Petroleum Bureau of China.
Experiments show that the preprocessing and character extraction strategies
are effective. The defects recognizing error rate is less than 0.5%. The result is
satisfying since the pig’s inspection precision is 10mm. 98% defects can be
assigned to correct levels.
We would like to acknowledge the State Key Laboratory of Ocean Engineering
for the opportunity to have experiments, and with a special thanks to Shengli
Petroleum Bureau offering correlative pipeline information and samples. This
work was supported by China National High-Tech Research and Development
Program #2001AA602021.
94
P1G-6
ULTRASONIC MONITORING OF PAPER COATING
BULK MODULUS DURING LAYER FORMATION.
T. KARPPINEN*, I. LASSILA, and E. HAEGGSTROM, Dept. Physical Sciences,
Univ. of Helsinki, Helsinki, Uusimaa, Finland.
Corresponding e-mail: [Link]@[Link]
We report on efforts to monitor, in real-time, the bulk modulus in the thickness-
direction of pigmented industrial grade paper coatings during layer formation.
The formation of this multiphase dispersion (latex-pigment-thickener-water)
determines largely the coating structure that has a significant impact on the
printability and the strength of paper.
The ultrasonic probing was conducted with a pitch-catch set-up. A broadband
20MHz ultrasonic longitudinal pulse propagated orthogonally towards the coating
through an aluminum slab on top of which coating paste was applied and reflected
from the aluminum-paste and coating-air interfaces. The longitudinal modulus
of the coating was calculated from the time-of-flight of the received signal and
the density of the coating that was obtained from a contemporaneous laser
distance measurement of the layer thickness.
Four different, 200 um thick, coatings were investigated in a controlled
environment. These were: 1) a typical coating consisting of calcium carbonate,
latex, and carboxy-methyl cellulose (CMC); and simplified versions of it, i.e. 2)
calcium carbonate and latex, 3) calcium carbonate and CMC, and 4) pure calcium
carbonate. The maximum bulk modulus for these sample kinds was 6 GPa.
This first and second critical concentration stages of the drying process were
identified.
Keywords: ultrasound, bulk modulus, longitudinal waves, pigment, paper coating,
nondestructive evaluation
We gratefully acknowledge Oy Keskuslaboratorio-Centrallaboratorium Ab for
providing us with samples.

P1G-7
HIGH PRECISION TUNING FORK SENSOR FOR
LIQUID PROPERTY MEASUREMENTS.
L. MATSIEV*, J. BENNETT, and O. KOLOSOV, Symyx Technologies, Santa Clara,
CA.
Corresponding e-mail: lmatsiev@[Link]
It was shown in [1] that if the piezoelectric resonator is immersed in a certain
liquid the changes in the resonator electrical impedance produced by a liquid
environment can be measured and used to independently calculate viscosity,
density, dielectric constant and conductivity of the liquid. The measurement
procedure described in [2], allows for precision of the data comparable to the
precision of reference analytical equipment, and can exceed the requirements
of ASTM D-445. The sensor based on this principle is currently being
commercialized for lubrication monitoring and other applications.

95
The equivalent circuit of a piezoelectric resonator with electrodes capacitance
and mechanical arms connected in parallel is a proven way of explaining the
interaction of such resonator with the environment. The fringing electric field
from the electrodes passes through surrounding liquid, which makes the
capacitance sensitive to the dielectric constant of the liquid. The mechanical
arm consists of the sequentially connected equivalent capacitance, resistance
and inductance. In [1] it was shown that for a flexural resonator the additional
impedance produced by surrounding liquid is connected sequentially to the
mechanical arm and the value is given by a sum of two terms, one of which is
proportional to the product of frequency and liquid density and the second one
is proportional to the square root of the product of frequency, viscosity and density.
Since these two impedance terms and the impedance of the electrodes
capacitance are orthogonal functions of the frequency, viscosity, density and
dielectric constant can be calculated independently from the total impedance of
the resonator measured at three different frequencies. In practice, it can be
advantageous to sweep the frequency within certain range and use the whole
impedance curve. Liquid properties are extracted from the curve by fitting it
using the least square method [3] with the liquid property values being free
variables. In this case the average error in the free variable can be related to the
error of impedance through an integral equation over the frequency range. The
impedance partial derivatives under the integrals serve as natural weighting
functions allowing for better averaging of the impedance errors acquired in the
course of frequency sweep. This method of the errors calculation helps analyzing
the contribution of various error sources: measurement system noise, non-
linearity, frequency instability, etc.
Experimental studies were performed using a quartz tuning fork and the technique
described above. The results of viscosity and density measurements for a set of
Cannon standard oils are shown in Table 1. The method demonstrated excellent
reproducibility (< 1%) in all samples the viscosities of which varied between 2
and 30 cP.
References:
1. L. Matsiev Application of Flexural Mechanical Resonators to High Throughput
Liquid Characterization. Proc. of 2000 IEEE Ultrasonics Symposium. v.1, 427-
434
2. L. Matsiev, J. Bennett, E. McFarland US Patent 6393895, 05/28/2002
3. O. Kolosov, L. Matsiev, M. Spitkovsky, V. Gammer US Patent Application
2004/0107055A1
Table 1. Peak-to-peak scatter in 10 sequential measurements at 50C°
CannonÔ StandardViscosity, cP Scatter, % Density, g/ccm Scatter, %

S3 2.008 0.352 0.8403 0.144

S6 3.757 0.407 0.8559 0.196
S20 10.65 0.187 0.8458 0.064
N35 18.25 0.298 0.8492 0.041
S60 29.72 0.256 0.8566 0.098

96
P1G-8
AN ULTRASONIC FLOWMETER IN PRODUCTION
BOREHOLES OF OILFIELDS.
H. ZHANG*1, W. LIN1, C. ZHANG1, D. WANG1, and T. LU2, 1Institute of Acoustics,
Chinese Academy of Sciences, Beijing, China, 2China Oilfield Services Limited,
Tianjing, China.
Corresponding e-mail: zhanghl@[Link]
In petroleum exploitation, it is desirable to monitor the flow in cased production
and injection boreholes for full borehole injection and production profiling. This
is especially true to have a picture of the flow variation with borehole depths,
from which the outflow from each layer of formations can be evaluated in
determining oil and water output production profiles. Traditional techniques for
the flow measurement, including ultrasonic flowmeters used for the flow
measurement inside a pipe, are not suitable for the situation. In this paper, a
prototype of an ultrasonic flowmeter used in a cased borehole is introduced.
The possibility to develop an effective ultrasonic flowmeter, similar to popular
well logging system, has been studied in the laboratory. The proposed tool is
made up mainly a steel cylinder pipe with a diameter in 60mm and a length in 1
meter, two ultrasonic transducers, and an electric circuit part. During the
measurement, the tool is put in a simulated borehole and kept on the borehole
axis. The ultrasonic pulse transmitted by one transducer is obliquely impinged
on the side wall of the borehole from which it is reflected several times and
eventually is received by the other transducer. The received signal is transmitted
through a cable to the ground for further processing. The time of the propagation
of the ultrasonic pulse is dependent on the velocity of the flow inside the borehole.
The relation between the time lag and the flow velocity in various conditions is
studied and is used in data analysis to extract the flow velocity. Both of the two
ultrasonic transducers act as transmitter and receiver alternatively and the date
for both of the propagation directions is compared to improve the measurement
precision. In our laboratory a system of 1 MHz is implemented and tested. The
precision is better than 10mm/s. The effect of the tool on the flow measurement
results is studied and a special treatment in reducing this effect is adopted in
the signal processing. In order to use this measurement system in real borehole
environments, all of the transducers and electric circuits are packed within a
thin pipe. This system is tested and it works well at the temperature up to 120°.

P1G-9
LAMB WAVE PROPAGATION IN A PIEZOELECTRIC
PLATE SUBJECTED TO CONDUCTIVE AND VISCOUS
FLUID LOADINGS.
S.-H. KUO* and Y.-C. LEE, Nation Cheng Kung University, Tainan,Taiwan.
Corresponding e-mail: n1890118@[Link]
Leaky Lamb wave is an interesting problem in the area of NDE and ultrasonic
sensor technology. When the lamb wave propagates on the plate surrounded
by a fluid, not only the mass loading effect but also the viscosity of the fluid will

97
cause the mode-shifting in the dispersing curve of leaky lamb wave. Furthermore,
when the plate is a piezoelectric material, the conductivity of the fluid also disturbs
the dispersion curve and this is known as dielectric loading effect. All these
three effects added up together make leaky lamb wave propagation a very
complicated problem in physical acoustics. In the last two decades, many people
have been investigated in this phenomenon and most of them have been using
a theoretical analysis method called partial wave theory. However, there are
very limited experiment results being reported in the literature and hence make
it difficult to verify this theoretical model.
In this paper, the leaky Lamb wave problem will be re-examined with theoretical
analysis as well as experimental measurements. First of all, the partial wave
theory is used to analyze the corresponding velocity shifting and attenuation of
leaky Lamb waves induced by the conductivity and viscosity. Secondly, a PVDF
lens-less acoustic microscopy measurement system is established to measure
the leaky Lamb waves using tone burst acoustic signals. It is basically an acoustic
interfering measurement system and can obtain V(z) curves at a selected
frequency. In comparison to the time-resolved or time-domain PVDF lens-less
acoustic microscopy, this tone burst measurement system can greatly improve
the accuracy in measurements of dispersion curves over a specific frequency
range.
Based on the V(z) curve measurements, the velocity shifting in leaky Lamb
wave can be accurately determined and then compared with theoretical data
calculated from partial wave theory. Good agreements between experimental
data and numerical results are observed, which verifies the analytic model of
the leaky Lamb wave. For a x-cut LiNbO3 sample plate, a significant dielectric
loading effect along a particular orientation has been identified by the partial
wave analysis and numerical calculation, and verified experimentally by the
V(z) measurement. This could be very useful for further development of dielectric
or conductivity sensors based on Lame wave propagation.
Key Words: Leaky Lamb Wave, Dispersion Curve, Fluid-Loading, Conductivity,
Viscosity, V(z) curve measurement

P1G-10
MECHANICAL CHARACTERIZATION OF TA AND TAN
DIFFUSION BARRIER LAYERS USING
LASER ACOUSTICS.
D. M. PROFUNSER*2,1, J. VOLLMANN1, and J. DUAL1, 1ETH Zurich, Center of
Mechanics, Zurich, Switzerland, 2Hokkaido University, Graduate School of
Engineering, Sapporo, Japan.
Corresponding e-mail: vollmann@[Link]
The usage of copper as replacement of aluminium in integrated circuits leads to
new technological challenges which are caused by its mechanical properties on
the one hand side and by is tendency to migrate into dielectric and/or
semiconducting layers on the other hand. To prevent such diffusion processes,
very thin layers consisting of tantalum and tantalum nitride are deposited.

98
A non-contact, non-destructive, short-pulse-laser-acoustic method is used to
determine the mechanical properties of the barrier layers and of the copper
layer. High frequency mechanical pulses are excited and detected using laser
pulses of 70 fs duration. For metals this leads to wavelengths of about 10 to 20
nm and the corresponding frequencies amount to 0.3 to 0.6 THz. Thin film
measurements of buried diffusion layers are provided and compared with
Scanning Electron Microscopy measurements (SEM) and Rutherford
Backscattering Spectroscopy measurements (RBS). Limits of the presented
method are discussed.
Keywords: Laser acoustic thin film metrology, Diffusion barrier measurement,
Photo acoustic detection, in-depth profiling

P1G-11
EVALUATION OF LOW-LIGHT ABSORPTION
MATERIAL BY PHOTOACOUSTIC MICROSCOPE.
A. MINAMIDE*1, N. NAOE1, and Y. TOKUNAGA2, 1Kanazawa Technical College,
Kanazawa, Japan, 2Kanazawa Institute of Technology, Ishikawa, Japan.
Corresponding e-mail: minamide@[Link]
A photoacoustic microscope can be used for the nondestructive evaluation and
imaging of cracks and defects in opaque materials. Unfortunately, conventional
photoacoustic microscopes are difficult to use for the evaluation of low absorption
materials, such as transparent materials and metal thin films, since these
materials do not form a strong heat source because they have high transparency
or high reflectivity for visible light.
In this paper, a new evaluation method for low-light absorption materials by a
photoacoustic microscope with low laser power is proposed. In the evaluation
of the transparent epoxy adhesive, the laser light is absorbed by an opaque
substrate under the adhesive, and a heat source is formed. A method to estimate
the thermal diffusivity of the adhesive that used the thermal wave from the heat
source is proposed.
In addition, attention is given to the remarkable behavior of the surface plasmon
generated on the boundary plane between metal and air as one possibility for
generating a strong heat source. If the surface plasmon forms on the boundary
plane of metal and air, it could be converted into heat for electron-phonon
interaction. Therefore, the surface plasmon may be used as a source of thermal
wave generation, a method that is called Surface Plasmon Thermal Wave
Microscopy (SPTWM). In this work, a basic system of SPTWM was constructed
and used in experiments that generated a strong heat source from the surface
plasmon in metal thin films. The Kretschmann configuration was used for the
excitation of the surface plasmon, and a microphone was used to measure the
photoacoustic signal generated from the heat source. The experimental results
obtained using a thin film of silver showed that the SPTWM was about 20 times
more effective than conventional methods for observing photoacoustic signals.
This work was partially supported by a Grant-in-Aid for Scientific Research from
the Japanese Ministry of Education, Culture, Sports, Science, and Technology.

99
P1G-12
ULTRASONIC TESTING OF THE TIME EVOLUTION
PROPERTIESOF ORANGES.
F. CAMARENA*, J. A. MARTÍNEZ-MORA, M. ARDID, J. RAMIS, and V.
ESPINOSA, EPSG-UPV, Gandía, Valencia, Spain.
Corresponding e-mail: fracafe@[Link]
Valencian region in Spain is one of the greater producers of citrus fruits of the
world. In this context, the capability to store them in good conditions is very
relevant. Classical measurements used in orange factories in order to obtain
information related to the quality of the fruit are dehydration and oil-gland break
stress, but those measurements are costly and destructive. The quality and
evolution with the storage time of the fruits properties inside the refrigeration
chambers can be tested by means of the use of ultrasounds. In this study we
report the techniques, experimental set up, measurement procedures and results
obtained to relate classical measurements (destructive) with the ultrasonic ones,
in order to test the possibility to design a non-destructive ultrasonic device able
to give information as turgidity and hydration of the oranges. Two different
ultrasonic experimental set-ups have been built: a destructive one to test the
properties of the orange skin and a non-destructive one to test the properties of
the whole orange. A sample of 200 oranges Navelina has been measured during
two months at normal conditions. Wave aluminium transformers have been used
to focalize the ultrasonic wave into the orange skin as well as for the subsequent
detection of the acoustic signal once crossed the peel. The design studies of
the transformers, the correct way to do the measurements, the effect of the
pressure, and the dispersion reached are described. Correlation between
classical, ultrasonic and impact test measurements has been obtained. We show
that the ultrasonic magnitudes change with the dehydration of the fruit and the
control procedure test can be carried out with ultrasound techniques. Finally, an
absolute calibration table allows us to estimate the hydration and turgidity of the
oranges using non-destructive ultrasonic techniques.
This study was supported by Generalitat Valenciana (GV04B/368) and Productos
Citrosol.

Session: P1H
ULTRASONIC MOTORS/LAB ON A CHIP
Chair: S. Bhave
Cornell University
P1H-1
A NEW SCHEME FOR EXPERIMENTAL-BASED
MODELING OF A TRAVELING WAVE
ULTRASONIC MOTOR.
H. MOJALLALI1, R. AMINI*2, R. IZADI-ZAMANABADI2, A. A. JALALI1, and J.
POSHTAN1, 1Iran University of Science and Technology, Tehran, Iran, 2Aalborg

100
University, Aalborg, Denmark.
Corresponding e-mail: mojallali@[Link]
In recent years, piezoelectric structures have been widely used in small scale
actuators specifically motors and new implementations have exhibited high torque
densities and similar efficiencies in comparison with conventional electro-
magnetic motors. As in the case of the traditional electromagnetic motors, the
experimental investigations, which highlight the behavior of the piezoelectric
motor, are used to derive the model of the motor as the first stage of speed and
position control.
In this paper, a new and high precision method for experimental modeling of
rotary piezoelectric ultrasonic motors by equivalent circuit approach is presented
in a thorough manner and examined on a Sashida USR60 piezoelectric motor
as the case study.
As the first step, an improved equivalent circuit model for free stator modeling of
traveling wave piezoelectric motor has been presented where the elements of
the circuit are complex. The circuit elements are derived through a new iterative
approach for determining the piezoelectrical material constants in the transverse
mode. The usage of complex values which considers the information of the
magnitude and phase of the admittance together, accounts the losses of the
piezoelectric vibrator with more precision. The new approach uses only the
measurements of five data points of admittance measurement. Comparison
between the obtained circuit model and real data illustrates the accuracy and
precision of the method.
In the next step, a model of the no load motor is derived through a new approach
by considering the correspondent effect between the free stator, rotor-stator
contact and rotor. A new method for compensating the effect of the rotor on
changing the resonance and anti-resonance frequencies and the overall
admittance of the un-loaded motor is presented which is based on altering the
complex values of the circuit elements of free stator model. At this stage, a
nonlinear parameter estimation method based on levenberg-marquardt algorithm
is used to find the best complex values for the equivalent circuit. Also a current
source is added to introduce the effect of the pressure of the rotor on the stator
and the resultant speed change. The value of this current source is determined
by using the theoretical relation between the tangential vibrting speed and the
rotary speed of the unload motor. It is shown that the admittance of this new
model agrees with the admittance measurements of the unloaded motor.
The paper continues by adding the model of the effect of the load. The loading
torque interfere on the model as a torque-dependant current source.
Simulation of the speed under various torques shows that this models meets
the results of the experiment.
Finally, the effect of the temperature on changing the resonance frequency of
the motor is modeled. The result of this simulation is shown and verified with the
experiments.
Comparison between the obtained circuit model and real data illustrates the
accuracy of the new equivalent circuit modeling method as an experimental
method for modeling of ultrasonic piezoelectric motors and specifically Sashida
USR60 as a case study.

101
P1H-2
TRAPPING AND EXTRACTION OF SMALL PARTICLES
BY VIBRATING SHARP EDGES.
J. HU*, J. YANG, J. XU, C. TAY, and Y. CAI, Nanyang Technological University,
Singapore.
Corresponding e-mail: ejhhu@[Link]
Manipulations of small particles have applications in biotechnology and many
other areas. In this report, we present the methods of trapping and extracting
small particles by ultrasonic actuators, which employ one or two tapered metal
plates operating at the flexural vibration mode. Small particles can be trapped
on the sharp edges of the metal plates. By the pumping effect in the gap between
two tapered metal plates in vibration, lighter particles can be extracted from a
mixture of particles. The capability of trapping and extracting small particles
was measured for different particles, electric input and vibration velocity. The
results show that the methods are effective in collecting particles in water and
air, and extracting particles from a mixture of particles in air.

P1H-3
FLOW-BIREFRINGENCE SPECTROSCOPY OF
MOLECULAR ORIENTATION WITH QUADRAPOLE
PIEZO-ACTUATOR SYSTEM.
M. HOSODA*1, K. HORII2, H. OGAWA1, K. TAKAGI2, H. NOMURA1, and K.
SAKAI2, 1Tokyo Denki University, Saitama, Japan, 2University of Tokyo, Tokyo,
Japan.
Corresponding e-mail: mhosoda@[Link]
We are interested in the physical properties of the soft condensed matters,
including those liquids of polymers, molecules of potential liquid crystals, and
self-assembling micelles. In particular, these liquids have different internal
degrees of freedom, and the coupling among them plays an important role in
the dynamics that characterizes the complex nature of the matter. Oscillatory
flow birefringence has been the common technique to observe the coupling
between the translational and rotational degrees of freedom of molecules.
However, the experimental difficulty in inducing the quickly alternating shear
flow of layered-mode has limited the frequency range of observation to 1kHz.
We recently developed a new spectroscopic technique of flow birefringence
induced with a set of four piezo-actuators in a quadrapole position surrounding
the sample chamber of 1mm square in size. One pair of them facing each other
is driven in a countering motion, while the other pair is also countering but in the
opposite phase. In this configuration, the liquid is activated into a periodical
shear deformation of completely pure mode without volume change. The induced
dynamic birefringence is observed by the optical ellipsometry technique. A laser
beam passing through the sample under the shear deformation suffers from the
modulation in its polarization, which is analyzed in the conventional way with
the wavelength plates. The ellipsometry signal thus obtained gives information
on the coupling between the molecular orientation and the shear deformation.

102
We chose the isotropic phase of 6CB as a typical sample that changes into
nematic phase below the critical point, and examined the performance of the
present system. The result showed that we could observe the coupling dynamics
between the shear flow and the orientation over the frequency up to 100 kHz.
The mechanical resonance of the actuators and their surroundings determines
this limiting frequency at present. A piezo-actuator system with smaller dimension
would improve the higher-frequency operation. The experimental system was
applied to the aqueous solutions of CTAB in the phase of wormlike micelle and
the spectroscopic study was made. We successfully observed the relaxation
effect of the orientational order at around 100 Hz, which agrees well with the
previous result obtained by the ultrasonic birefringence method.

P1H-4
ULTRASONIC MMICROMANIPULATION OF LIQUID
DROPLETS FOR A LAB-ON-A-CHIP.
M. TAKEUCHI* and K. NAKANO, Faculty of Engineering, Tamagawa University,
Machida, Tokyo, Japan.
Corresponding e-mail: takeuchi@[Link]
Noncontact and nondestructive techniques for manipulation of micron-sized
particles in liquids, such as transportation, trapping, and separation, are of
considerable interest for biotechnology and micromachining. Previously we have
proposed ultrasonic micromanipulation (UMM) techniques based on the radiation
forces of very-high-frequency (VHF) ultrasound. Recently, with the emergence
of the so called Lab-on-a-Chip (LOC), various chemical processes such as
transport, reaction, separation, purification and detection are now integrated on
glass and plastic substrates of with dimensions of several centimeters. Further
development of LOC requires new microfluidic technology to process liquid
droplets of a very small size (0.5~60µl) without requiring complicated pumps
and tubes. In this paper, we report two types of ultrasonic micromanipulators for
use in liquid droplet processing. The first type of manipulator uses a single
phase unidirectional interdigital transducer on a piezoelectric Lamb waveguide.
The second type of manipulator uses wedge type transducers on an isotropic
glass plate. Experiments using a frequency of about 6MHz confirm one-
dimensional and two-dimensional transportation of water droplets. Experimental
results are given for the relationship between the moving speed and the volume
of the water droplet. Also presented is work in progress on the development of
visual feedback techniques.

P1H-5
SMART SEPARATION DEVICES FOR PARTICLE
CONCENTRATION IN WATER USING ULTRASONIC
STANDING WAVE.
Y.-S. LEE* and J. KWON, Smart Measurement Group, Korea Research Institute
of Standards and Science, Daejeon, South Korea.
Corresponding e-mail: highfield@[Link]

103
This paper presents the theory, design, manufacture, experiment and evaluation
of smart devices for enhanced separability of tiny particles in water. The smart
devices take advantage of the ultrasonic standing wave, which was generated
by the operation of piezoceramic PZT patches installed on the separation devices.
Since the devices were developed for the concentration of plant cells, the design
and manufacture of them were targetted to the improvement of conventional
separation processes.
The details of the device design including the electro-acoustical modelling for
separation and PZT transducers are described at first. Two different ultarsonic
wavelengths in half (1/2) for a nodal line in the middle and quarter (1/4) for a
nodal line at the boundary of each separation channel respectively were applied
in the design to compare their separation performance of the cells. Each
separation channel has 370 micrometers width with the operating frequencies
of 1 MHz and 2 MHz. Based on this design, the separation devices were analyzed
with finite element approach to confirm the design parameters and to predict
the acoustic pressure field in the separation channels of the devices.
Results of the experiment showed that the quarter wavelength in the separation
channel gave a better performance in concentration of plant cells rather than
the half wavelength one. Also the electric impedances of the PZT patches bonded
on the separation devices were measured to compare with theoretical analysis.

P1H-6
FREE STATOR ANALYTICAL MODELING OF A NOVEL
LOW VOLTAGE MULTILAYER
PIEZOELECTRIC MOTOR.
H. MOJALLALI1, R. AMINI*2, R. IZADI-ZAMANABADI2, A. A. JALALI1, and J.
POSHTAN1, 1Iran University of Science and Technology, Tehran, Iran, 2Aalborg
University, Aalborg, Denmark.
Corresponding e-mail: mojallali@[Link]
In recent years, many research activities in the field of ultrasonic motors and
piezoelectric actuators have been done because of their attractive features.
Most of developed ultrasonic motors operate at a high voltage which is a demerit
in the application areas of ultrasonic motors.
In this paper, at first we introduce a novel and small rotary piezoelectric ultrasonic
motor based on the well known traveling wave principle, the dimensions of which
are 15.0 mm in diameter and 5.0 mm in thickness. The usage of multilayer
ceramic in the stator structure of the piezoelectric motor has led to low drive
voltage, quick response time, large generative force and high electromechanical
coupling.
Then, a novel analytical model, which deals with the electromechanical behavior
of multilayer structure of free stator, is derived. In this model, a new form for the
constitutive equations of multilayered structure of stator is presented where the
electromechanical coupling is as an additional stiffness matrix. This matrix shows
properties of the piezoelectric structure and is not dependent on mechanical
boundary conditions. But some of the parameters in this analytical model such
as stiffness, modal mass and damping coefficients should be determined through

104
experiments on the multilayer structure. This identification performs by the aid
of an equivalent circuit model of the free stator which is recently introduced in
our previous published works. This hybrid approach for determining the free
stator unknown parameters results a high precision model from the combination
of the analytical and experimental modeling approaches.
Finally, the validity and accuracy of the free stator model is verified by simulation
and comparison to the real data obtained from the multilayer rotary piezoelectric
motor.

P1H-7
DEVELOPMENT OF GYRO-MOMENT MOTOR.
Y. TOMIKAWA*, T. TAKANO, and M. AOYAGI, Yamagata University, Yamagata
Japan.
Corresponding e-mail: tomikawa@[Link]
This paper deals with developmental research of the gyro-moment motor. This
motor is a construction of a pizoelectric vibratory motor, which was named a
gyro-moment motor by one of authors([Link]). The gyro-moment motor
was composed of a stator vibrator of a thin rectangular plate and a T-type rotor
set on the center point of the plate stator vibrator. The stator vibrator was
electrically driven by the piezo-ceramics bonded on it. As mentioned above, this
vibratory motor was of a new concept which nobody might think of before. In our
previous papers(1), (2), the fundamental principle and some measured
characteristics of a such motor were reported to show that the rotor of T-type
was merely made to rotate by such a motor construction and what sizes of
rotors should be chosen to rotate with a high speed. However, the gyro-moment
motor of a trial production was a littel big; that is, sizes of the stator vibrator were
160mm(length)x30mm(width)x0.5mm(thickness), which might not be suitable
for a practical use. Therefore, after previous works, we have further developed
the gyro-moment motor in order to make clear its operation and to improve
weak points of its characterisitics through experiments. That is, in this paper, we
would like to present the experimental results.
Main points of the experimental works of this timeare followings: 1) A smaller
sized motor construction than the previous motor was dealt with , 2) A self-
oscilated-type gyro-moment motor was tried, because the previous motors
needed an external punching-force in its starting of rotation.
The experimental results proved that this problem could be solved by somewhat
modification of T-type rotor. 3) As another motor construction, we also devised
the gyro-moment motor which can be called a stator-vertically-set type. It was
found clear that this motor might be applied as a kind of toy. We think this will be
useful for physical excercise of senior persons, too.
References :
(1) K. Kanauchi and [Link]; “Gyro-Moment Motor”, ACTUATOR 2000 19-
21 June 2000, Bremen, Germany No.B2.2 pp.246-249.
(2) [Link] and [Link]; “Characterisitcs of Gyro-moment motor”,
ACTUATOR 2002, 10-12 June 2002, Bremen, Germany, No.P15 pp.450-453.

105
Session: P1I
PIEZOELECTRICAL TRANSFORMERS/MOTORS
Chair: M. Kurosawa
Tokyo Institute of Technology
P1I-1
EFFECTS OF ELECTRIC LOAD ON VIBRATION
DISTRIBUTION OF A PIEZOELECTRIC TRANSFORMER
OPERATING AT THE THICKNESS SHEAR MODE.
J. DU*, J. HU, and K. J. TSENG, Nanyang Technological University, Shcool of
EEE, Singapore.
Corresponding e-mail: djl@[Link]
In this study, a theoretical model for analyzing the effect of electric load on the
vibration distribution of a piezoelectric transformer operating at the thickness
shear mode was developed. The relationship between the vibration distribution
and load resistance of the transformer was analyzed by this model. The vibration
of output part decreases along the direction of vibration transmission. The larger
the difference between the load resistance and the matching load is, the more
the vibration of output part decreases. The theoretical conclusions were justified
by experiment measurement and FEM analyses. The model provides guidelines
of optimizing the transformer and explains some important phenomena in
piezoelectric transformers, such as a higher temperature at the input part of
almost all piezoelectric transformers.
The authors would like to acknowledge the financial support by Academic
Research Fund RG4/102 from the Ministry of Education, Singapore.

P1I-2
MODELING AND CHARACTERISTICS
COMPARISON OF TWO DIFFERENT
PIEZOELECTRIC TRANSFORMERS.
J. M. FERNANDEZ* and Y. PERRIARD, Ecole Polytechnique Fédérale de
Lausanne, Lausanne, Switzerland.
Corresponding e-mail: [Link]@[Link]
Recent progress on electrical power applications, particularly in the domain of
the miniaturization of both electrical and electromechanical devices, has enabled
envisaging the development of very compact systems (microcomputers,
microsystems, cellular telephones, etc.). The supply, storage and transformation
of electrical power in these systems must respect this trend. However, the
miniaturization of classical electromagnetic transformers raises certain problems
such as manufacture both the coils and the magnetic core, increase in magnetic
leak and especially electromagnetic pollution of the environment. One interesting
solution consists of using a piezoelectric transformer. This will insure an electrical-
mechanical and mechanical-electrical double conversion of energy with a
transformation ratio that allows adapting the output voltage being used. This

106
kind of transformer, which is more compact, lends itself better to miniaturization,
in addition to displaying the attractive characteristics of immunity to the magnetic
field. In this paper a Rosen-type structure of piezoelectric transformer has been
initially studied. The first approach consists of performing an analytical model.
This approach is inspired by the technique used in electroacoustics and allows
to build an electric equivalent circuit of the vibrating structure. The second
approach deals with numerical finite element method using ANSYS software.
The 3D structure has been simulated and the validity of the hypotheses done in
the equivalent circuit method has been proved. In order to validate the models,
experiments on two different multi-layer piezoelectric transformers, one from
Noliac and the other from Transoner have been performed. Measures of the
global transformation ratio versus frequency under different values of the load
resistance have been obtained. As an example, for a load resistance of 200kΩ,
the voltage transfer ratio is about 100 at the resonance frequency (132kHz) for
the Noliac type and about 80 (at 66kHz) for the Transoner type. With regard to
the output power, it varies from 1.5W in the first case to 5W for the second one.
Then, a piezoelectric transformer using radial extensional vibration mode has
been studied and characterized. An analytical model has been investigated using
linear piezoelectric equations combined with the theory of elasticity, and an
equivalent circuit model has been deduced. A three-dimensional finite element
model has been then simulated using ANSYS to find out the resonance frequency
but also to determine the transformation ratio as a function of the frequency for
several values of the load resistance. The models are then validated through
experimental results using a Transoner radial type piezoelectric transformer.
Finally, voltage gain, output power and efficiency of both piezoelectric transformer
types are compared taking into account their characteristics. This will allow a
case study depending on the application, especially in the pre-study process,
where one of these two different piezoelectric transformers is to be chosen.

P1I-3
A POWER ULTRASONIC ACTUATOR BASED ON A
DISPLACEMENT AMPLIFIER VIBRATING
IN FLEXURAL MODE.
A. IULA*1, L. PARENTI1, N. LAMBERTI2, and M. PAPPALARDO1, 1University
Roma Tre, Roma, Italy, 2University of Salerno, Salerno, Italy.
Corresponding e-mail: iula@[Link]
Commonly used high displacement ultrasonic actuators are composed of a power
vibration generator (the Langevin transducer) and of a displacement amplifier.
This structure is employed in a large variety of industrial applications, as well as
in the biomedical field as an ultrasonic bystoury.
Among possible displacement amplifiers, sectional ultrasonic concentrators,
made from rods of variable and constant cross section, are those that have
been mainly exploited in applications. Basically, sectional concentrators are
designed to resonate in length-extensional mode at the same frequency of the
Langevin transducer and the displacement amplification depends on the ratio
between the back and the front sections.

107
In this work a new type of ultrasonic actuator has been proposed. It is composed
of a symmetrical Langevin transducer working in a length-extensional mode
joined to a displacement amplifier that is designed to vibrate in flexural mode.
The displacement amplifier is composed of: a cylinder shaped base with the
same radius of the Langevin transducer, of four arms, and of a thin disk acting
as a displacement collector.
This kind of structure is able to transform the almost flat displacement provided
by the Langevin transducer into a flexural deformation of the arms. This
deformation is transmitted to the collector. In this way, an axial symmetric flexural
mode can be excited in the collector. In this case, the maximum value of the
axial displacement occurs at the centre of the collector.
In analogy to classical actuators that use sectional concentrators, the Langevin
transducer and the flexural amplifier have been separately designed to work at
the same frequency. Design and analysis of the actuator have been performed
by using a FEM commercial package (ANSYS).
Performances of the proposed actuator have been compared with that of a
classical stepped horn actuator designed to work at the same frequency (about
50 kHz) and with the front section equal to the total section of the areas of the
arms.
Simulated results have shown that the proposed concentrator exhibits a
displacement amplification about 50 % higher than the stepped horn.
Furthermore, by joining the two concentrators to the same Langevin transducer,
and driving the two actuators with the same current amplitude, the proposed
actuator is able to absorb a higher electrical power than the classical one.
Due to superposition of these two effects, the proposed actuator has shown a
maximum displacement that is about twice the maximum displacement of the
classical actuator.
In order to validate numerical results, two prototypes have been manufactured
by following FE design. Measurements of the maximum displacement carried
out on the two actuators are in a good agreement with simulation results, proving
that the proposed actuator is more suitable than the classical stepped horn in
application where high displacements are required. As an example, due to the
flexural deformation of the front surface, the proposed actuator could be exploited
as a mirror with variable bend radius to deflect laser beams in optical applications.

P1I-4
AN ULTRASONICALLY LEVITATED NON-CONTACT
SLIDING TABLE WITH THE TRAVELING VIBRATIONS
ON FINE-CERAMIC BEAMS.
D. KOYAMA*, T. IDE, J. FREND, K. NAKAMURA, and S. UEHA, Precision and
Intelligence Laboratory, Tokyo Institute of Technology, Yokohama, Japan.
Corresponding e-mail: dkoyama@[Link]
This report presents a new design and its performance test for a non-contact
sliding table by ultrasonic levitation. A slider is levitated by an acoustic radiation

108
force emitted from two vibrating guide rails, and a flexural traveling wave
propagated along the rail guides allows a non-contact transportation of the slider.
The reduction of the abrasion and dust generation, high cost performance and
simple structure are expected. The profile of the sliding table was designed
using the finite element method (FEM) for high levitation and transportation
efficiency. The twist mode and the in-phase flexural mode were selected and
tested.
The prototype sliding table made of alumina ceramics for obtaining higher
machining accuracy and rigidity. A structure consists of a pair of 152-mm-long
beams with a triangle cross-section, two bolt-clamped Langevin transducers
with the diameter of 11 mm and cross-shaped vibration direction converters to
permit low height of 11 mm. A slider with the length of 40 mm was designed to fit
the two rail guides.
The vibration modes of the guide rail were investigated with a laser Doppler
vibrometer. The experimental results showed that the modes designed by FEM
were successfully excited near the expected. The flexural mode standing wave
was observed along the guide rail at the resonance frequencies of 89.4 kHz,
and the levitation of the slider could be confirmed even if the levitation distance
is less than 10 µm. The levitation distance of the slider was measured with
increasing the slider’s weight, and the levitation force, the levitation rigidity and
the displacement amplitude of the levitating slider in the vertical direction were
measured to be 4.8 kN/m2, 2.5 kN/µm/m2 and under ±1 µm, respectively, in the
case of the levitation distance of 2.2 µm.
The non-contact transporting of the slider was achieved with the phased drive
difference of two transducers. By controlling the phase difference, the slider
transportation direction could be switched. The maximum thrust and the
transportation speed were 1.3 mN and 34.6 mm/s, respectively, in the case of
the slider’s weight of 107 g.

P1I-5
MINI-DRIVER FOR ULTRASONIC MOTOR
BASED ON CPLD.
L.. HUAFENG* and Z. CHUNSHENG, Nanjing Univ. of Aeronautics and
Astronautics, Nanjing, China.
Corresponding e-mail: lihuaf@[Link]
As a novel motor, UltraSonic Motors (USM) exhibit advantages over conventional
electromagnetic motors. For example, USM can produce a relative high torque
at a low speed with a high efficiency, and the torque produced per unit weight is
high. These features are useful for utilizing as gearless actuators or direct servo
drives. The motors have recently been applied as direct drive actuators for
articulated robots, actuators for control valves and a positioning table of machine
tools because they require quick response and precise position control of
actuators. Some experts even predict that USMs will replace micro electro-
magnetic motors in certain special areas in the future.
But in order to drive the USM, a special driver is required, which has been an
obstacle for replacement of traditional motors by USMs. If the driver has a big

109
volume, the promotion of USM would be more difficult. Therefore, on the premise
of meeting the basic requirements, the volume of the driver must be reduced to
the greatest extent, so as to exploit the particular advantages of USMs in more
areas.
In order to fulfill this command, a new CPLD based mini-driver for ultrasonic
motor is presented in this work. Since the square-wave outputted by transformer
is filter to sine-wave to drive the USM, there must be a most suitable duty cycle.
So, the relation between the output voltage and energy efficiency of ultrasonic
motor driver and the duty cycle of driving pulse is analyzed in details firstly. The
rationality of 37% duty cycle is proved based on the analysis. Then a mini-driver
based on CPLD for ultrasonic motor is designed and manufactured. The driver
is applied to a TRUM-45 type traveling-wave ultrasonic motor, which is developed
by our Research Center. Experimental results show that this scheme is valid
and feasible. Compared with original driver composed by separate logical circuit,
this mini-driver can be reduced by 66% in volume and 40% in element numbers,
which fulfils the demand of practicability for general engineering.
This work is sponsored by Key Project of National Natural Science Foundation
of China (Project No.50235010)

P1I-6
DEVELOPMENT OF A MINIATURE ULTRASONIC
MOTOR USING A HELICAL COIL AS A STATOR.
T. MORIYA* 1, Y. FURUKAWA 2, Y. AKANO 1, and A. NAKAJIMA 2, 1Tokyo
Metropolitan University, Hachioji-shi, Tokyo, Japan, 2Tokyo University of
Agriculture & Technology, Koganei-shi, Tokyo, Japan.
Corresponding e-mail: moriya@[Link]
Preliminary results for the development of a miniature ultrasonic motor using a
helical coil as a stator are described. The coil was constructed using an acoustic
waveguide such as stainless steel wire. A rotor can be placed in or outside of
the coil. An ultrasonic transducer was attached at the end of the waveguide.
The basic principle of the motor is that of the traveling wave type ultrasonic
motor. That is, when the Lamb waves were transmitted along the coil, the waves
cause the elliptical motion on the surface of the waveguide, which is transmitted
to the rotor.
The group and phase velocities of the Lamb wave propagating along the
waveguide were analyzed using PZFLEX, and compared with the experiments.
The waveguides with a rectangular cross section ranging from 0.1 mm x 0.1
mm to 1 mm x 1 mm were used for the analysis. The experiments were conducted
using stainless steel wires that have a circular cross section. Reasonable
agreements between the simulations and the experiments were obtained.
The principle of the motor was confirmed experimentally. The wire diameter, the
number of turns, the height and radius of the coil were 0.3 mm, 7, 3 mm, 1.1
mm, respectively. A stainless steel wire with 0.5 mm in diameter was used as a
rotor. The motor was operated at 50 kHz. The rotation frequency ranged
approximately from 500 rpm to 1000 rpm according to the applied ultrasonic
power. The torque was estimated to be less than 100 nN-m. The motor worked
at 200 kHz, and 400 kHz.
110
parallel matching inductance. The deviation between simulated and measured
centre frequency was found to be 0.11 %.
References
[1] L. Reindl et al., “SAW Device as Wireless Passive Sensors”, Proc. IEEE
Ultrason. Symp. 1996, pp. 363-367
[2] A. Pohl, “A Review of Wireless SAW Sensors”, IEEE Trans. Ultrason.,
Ferroelect., Freq. Contr., vol. 47, no. 2, 2000, pp. 317-332
[3] C. W. Ruppel et al., “Review of Models for Low-Loss Filter Design and
Applications”, Proc. IEEE Ultrason. Symp. 1994, pp. 313-324
[4] R. F. Milsom, “Analysis of Generation and Detection of Surface and Bulk
Acoustic Waves by Interdigital Transducers”, IEEE Trans. Sonics Ultrason., vol.
SU-24, no. 3, 1977, pp. 147-166
[5] A. R. Baghai-Wadji et al., “Analysis and Measurement of Transducer End
Radiation in SAW Filters on Strongly Coupling Substrates”, IEEE Trans.
Microwave Theory Tech., vol. 37, no. 1, 1989, pp. 150-158
[6] K. Ibata et al., “Polynomial Expressions for SAW Reflection by Aluminium
Gratings on 128°YX-LiNbO3”, Proc. IEEE Ultrason. Symp. 1998, pp. 193-197

P1J-3
CLOSED-FORM COM MODELING OF REFLECTIVE
SAW TRANSDUCERS WITH ARBITRARY POLARITY
SEQUENCE AND APODIZATION.
A. S. RUKHLENKO*, Institute of Microtechnology, Neuchatel, Switzerland.
Corresponding e-mail: rukhlenko@[Link]
A new closed-form Coupling-of-Modes (COM) model of reflective polarity-
weighted and/or apodized surface acoustic wave (SAW) transducers has been
developed. The results are presented in the form of the complete mixed scattering
matrix comprising the acoustoelectric and electroacoustic conversion functions
and transducer admittance (radiation conductance and susceptance). The
solution is found in terms of the normal modes in the infinite periodic grating
which are two counter-propagating waves. Acoustic fields are described by the
coupled fundamental and the first backward spatial harmonics, with the additional
terms accounting for the electrical excitation of the interdigital array. The normal
modes of the infinite array are superimposed to satisfy the boundary conditions
at the SAW transducer ends (acoustic ports). Two important model parameters
are to be determined a priori: 1) wave propagation constant in the grating; 2) the
coupling factor between the fundamental and the first backward spatial
harmonics. The former is found from the dispersion equation for the short-circuit
infinite grating. The latter is related to the reflection factor of the semi-infinite
periodic grating. Given a strip reflection coefficient, the coupling factor is deduced
in the closed-form by applying the Morgan′s reflective array model equation to
the extreme case of the periodic array with the infinite number of the strips. The
classical COM equations follow from the theory in the particular case of a solid-
finger SAW transducer with a regular sign alternation. The equations are general

113
comprising reflective apodized SAW transducers. They allow to accurately predict
SAW reflections at the transducers ends even for split-finger apodized SAW
transducers. In the case of reflectionless SAW transducers, the theory is reduced
to the regular quasi-static approximation equations. The analysis results for solid-
finger withdrawal-weighted SAW transducers are presented which agree well
with the published results (reflective array model) and experimental data.

P1J-4
EMPIRICAL EQUATION FOR THE TEMPERATURE
COEFFICIENT OF FREQUENCY DERIVED FOR LEAKY
SAW DEVICES.
S. SABAH*1, E. CHILLA1, and D.- P. CHEN2, 1VI Telefilter, Teltow, Brandenburg,
Germany, 2Vectron International, Hudson, NH.
Corresponding e-mail: sabah@[Link]
Temperature dependence has been proved to be a significant design parameter
for the development of SAW devices. It is well known that many critical
parameters are temperature dependent, e.g. velocity, reflection coefficient,
transduction coefficient, electro-static capacitance and resistance.
Thus the performance of SAW filters, in particularly centre frequency, insertion
loss, group delay, input and output impedance, varies with temperature. However,
the temperature behavior is even more complicated since it is not only dependent
on material constants but also on the filter structure itself, e.g. type of
configuration, film thickness to wavelength ratio, mark to period ratio, as well as
the wave (Rayleigh wave, leaky SAW, STW) and mode (transverse or longitudinal
modes) utilized.
This paper reports on the determination of the temperature coefficient of
frequency (TCF). We present experimental data of leaky materials like LiTaO3
and LiNbO3 and derive an empirical equation for TCF. The agreement between
the measured and the calculated values is within 2 ppm/K. We found that the
TCF is mainly dependent on the film thickness to wavelength ratio and compared
the measured data with numerical calculations.

P1J-5
TRANSMISSION AND REFLECTION COEFFICIENTS OF
SAW IN BASIC CELLS OF SPUDT.
R. TAZIEV*, Institute of Semiconductor Physics of Russian Academy of Sciences,
Novosibirsk, Novosibirsk, Russia.
Corresponding e-mail: taziev@[Link]
In the previous paper [1] we suggested a scanning window technique in SPUDT
filter optimization, where we used 15 different basic cells for construction of
SPUDT filter structure. The purpose of the presented paper is to simulate
accurately transmission and reflection coefficients of SAW, taking into account
bulk wave losses, in the basic cells of such irregular, nonperiodic structure of

114
SPUDT by rigorous combined finite element and boundary element (FEM/BEM)
method. For every basic cell, which consists of 3 or 4 electrodes with different
width and finite height, electrically grounded or open-circuited, we simulate
transmission and reflection coefficients of SAW and bulk wave losses. That
allows us to evaluate P-matrix parameters of every cell properly, and compute
the characteristics of SPUDT on YZ LiNbO3. The results obtained were compared
with exact analysis of the same SPUDT structure by rigorous FEM/BEM software,
similar to [2].
[1] [Link], [Link] and [Link]. Scanning window technique in SPUDT
optimization, 2004 IEEE Ultrasonics Symposium Proceedings, pp.145-149.
[2] P. Ventura, [Link] and B. Lopes. Rigorous analysis of finite SAW devices
with arbitrary electrode geometries, 1995 IEEE Ultrasonics Symposium
Poceedings,pp.257-262.

P1J-6
ANALYSIS OF SAW DEVICES USING FEM/BEM
METHOD AND PARALLEL COMPUTING.
X. PEROIS*, T. PASTUREAUD, P.-A. GIRARD, and R. LARDAT, Temex, Sophia
Antipolis, Fance.
Corresponding e-mail: [Link]@[Link]
The main work for the SAW designer is to improve the filter performances,
keeping simulation and prototyping time as short as possible. In order to avoid
costly and time-consuming technological iterations, a very accurate simulation
tool is required.
Finite transducer FEM/BEM models give very accurate predictions for
synchronous or asynchronous structures like resonators or DMS on various
piezoelectric materials. A wide range of phenomena is taken into account through
the Green function and the Finite Elements formulation: mechanical effects like
Leaky SAW conversion into bulk waves or mechanical losses in materials
(electrodes or piezoelectric), or electrical effects like resistive losses in the
electrodes or ground plane influence through a special treatment of the Green
function.
The drawback of this approach is a very time and memory consuming model
that can be incompatible with short design time and cost effective computing
system. In order to reduce computation time and extend memory capacity, the
simulation software was modified for parallel computing using Message Passing
Interface standard and run on a multiprocessors PC’s cluster. Two different ways
were used to optimize the parallel software performances. The first one shares
the large FEM/BEM matrix over the memory of the computation nodes, allowing
to solve large problems, but the network communications limits the computation
time performance. The second one distributes every frequency computation,
allowing a quasi-linear acceleration of the computation time, but the FEM/BEM
matrix must be small enough for the memory of one computation node.
[1] P. Ventura, J.M. Hodé, M. Solal, J. desbois [Link], “Numerical methods for
SAW propagation characterization”,1998 IEEE Ultrasonics Symp. proc., pp 175-
186

115
[2] N. Finger, G. Kovacs, J Schöberl, U. Langer, „Accurate FEM/BEM simulation
of SAW filters“, 2004 . IEEE Ultrasonics Symp. proc
The authors are grateful to A. Mazari and N. Béreux-Vialle (CMAP, Palaiseau),
for the software development and help-ful support on the cluster configuration

P1J-7
FINITE-DIFFERENCE TIME-DOMAIN SIMULATION OF
DISPERSIVE LAYERED SAW FILTERS INCLUDING
ELECTRODE MASSLOADING.
K. Y. WONG* and W. Y. TAM, The Hong Kong Polytechnic University, Hong
Kong.
Corresponding e-mail: [Link]@[Link]
A number of analytical and numerical techniques have been developed to
simulate the frequency response of a SAW filter. However, due to the future
wireless communication applications, the present modelling schemes are difficult
to accurately predict the characteristics of SAW filters with complex interdigital
transducer (IDT) and fabricated on multilayered substrates. Furthermore, none
of them except the direct finite element method (FEM) can accurately predict
the performance of SAW filters for high frequency applications when second
order effects become significant.
Our goal in this paper is to extend the finite-difference time-domain (FDTD)
technique to analyze the frequency response of dispersive layered surface
acoustic wave (SAW) filters. The massloading effects of the IDT are also
considered. In this method, the partial derivatives of quasi-static Maxwell′s
equations and equation of motion are discretized to centered finite-differences.
Furthermore, the perfectly matched layer (PML) boundary condition is applied
to reduce the spurious reflections.
The main advantages of this technique are that the characteristics of the wave
propagating in inhomogeneous and non-linear media can be calculated.
Furthermore, with FDTD, specifying a new structure to be modelled is reduced
to a problem of mesh generation rather than the potentially complex reformulation
of an integral equation. Moreover, the update algorithm in the FDTD method is
efficient because it is based on the past field values at nearest neighbour
components, and unlike other numerical techniques such as FEM, requires no
matrix inversion. This technique is particularly suitable for the broadband analysis
because the frequency response of the wide frequency band can be computed
from a wideband Gaussian impulse excitation. In addition, FDTD method
generating time-marched arrays of field quantities are suitable for using the
computer visualization to illustrate the field dynamics1,2.
In this paper, the proposed model is applied to analyze the frequency response
of a ZnO/Diamond/Si layered SAW filter including the massloading effects of
IDT. The wave-front images of the acoustic waves are simulated and the phase
velocities of the Rayleigh and 1st Sezawa waves are extracted from the time
domain responses by using the Prony′s method. The simulated results are in
good agreement with the existing experimental data, indicating that the FDTD
method is an appropriate approach for modeling layered SAW filters.

116
Index Terms - FDTD methods, surface acoustic wave filters and frequency
response.
1
A. Taflove, and S.C. Hagness, Computational electrodynamics: the finite-
difference time-domain method,2nd Edition, Artech House, 2000.
2
P.M. Smith, and Wei Ren,′Finite-difference time domain techniques for SAW
device analysis,′ in IEEE Ultrasonics Symp., vol. 1, Oct. 2002, pp. 325-328.
The work described in this paper was partially supported by a grant from the
Research Grants Council of the Hong Kong Administrative Region, China (Project
No. PolyU 5135/02E).

P1J-8
FAST AND PRECISE MULTI-PURPOSE P-MATRIX
SIMULATION TOOL: BUILT-IN FEATURES AND
APPLICATION EXAMPLES.
R. LARDAT*, T. PASTUREAUD, and W. STEICHEN, TEMEX.
Corresponding e-mail: [Link]@[Link]
The need for time to market products has increased due to the reduction of the
market’s growth for SAW components in RF telecommunication. To reach the
challenging goal of designing SAW filters quickly with good characteristics, one
needs a fast and precise filter response simulation tool. This tool can then be
used inside an optimisation loop to achieve customer’s requirements in a short
period of time.
The aim of the presented tool is to simulate the following phenomena: fast and
precise capacitive computation, ohmic losses, P-Matrix dispersive parameters,
SAW directivity, multi-modes simulation used for transverse modes and
continuum.
It is based on P-Matrix model [1] coupled with a fast electrostatic solver that
computes the capacitive admittance taking into account for the real geometry
and polarities of the filter. This 2D BEM solver may use semi-infinite substrate
or can take into account for mass planes either on back surface or above active
area representing the package metallic lid. This 2D BEM solver uses spatial
electrostatic Green function and Chebishev polynomial for a reduced number of
unknowns (typically only 3 dof per electrode). For very large SAW devices, one
can accelerate the resolution by using “windowing” technique over a limited
number of electrodes.
Ohmic losses are modelled by adding a serial resistance at each electrode
connecting point. This leads to the construction of a linear problem whose size
is the number of electrodes. Exact resolution of such problem is time consuming
and an approached solution can be found using the knowledge of the capacitive
mutual admittance.
The P-Matrix parameters take into account for directivity in addition to usual
phenomena (SAW velocity under the grating, reflection coefficient, propagation
losses, coupling factor) and they may be frequency dependant for simulating
dispersive phenomena [2] which is well adapted for SAW devices using Pseudo
SAW.
117
The P-Matrix model may have an unlimited number of both electrical and
acoustical ports. This enables, for instance, the simulation of transverse modes
and continuum as an approach for diffraction [3].
In addition to filter admittance matrix, this tool may produce the charge distribution
due to capacitive coupling and the acoustic power representation along the
simulated device. The latter option enables an easy understanding of longitudinal
acoustical modes present, for instance, in Double Modes SAW devices.
Examples of various applications will be given: filters on Tantalate 42 and Niobate
41 for dispersive PSAW, high Q resonators on Quartz ST with slanted propagation
vs. X axis showing directivity and guided modes.
[1] J.M. Hode et al, “SPUDT- Based filters: Design Principles and Optimization”,
1995 IEEE Ultra. Symp.
[2] T. Pastureaud, “Evaluation of the P-Matrix parameters frequency variation
using periodic FEM/BEM analysis”, 2004 IEEE Ultra. Symp.
[3] M. Solal et al, “A P-Matrix based Model for SAW Grating Waveguides taking
into account Modes Conversion at the Reflexion”, 2003 IEEE Ultra. Symp.

P1J-9
COM MODEL OF THE IMPEDANCE-LOADED
SAW SENSOR.
Q. FU*, W.-J. FISCHER, and H. STAB, Dresden University of Technology,
Dresden, Sachsen, Germany.
Corresponding e-mail: fu@[Link]
The wirelessly requestable RF passive impedance-loaded surface acoustic wave
(SAW) sensor consists of one transmitting/receiving inter-digital transducer (IDT),
one or more programmable reflectors and a classical sensor connected to one
of the reflectors. In the last decade, many researchers have made great efforts
researching such devices and succeeded in simulating impedance-loaded SAW
sensors based on scattering parameters of a realized SAW delay line measured
by a vector network analyzer. In this work, the coupling-of-modes (COM) model
of SAW IDT is used to form a model of the RF impedance-loaded SAW sensor,
which does not require any measured data. This model is suitable for designing
and optimizing impedance-loaded SAW sensors, because it consumes less
computing time and it is much easier to obtain the relationship between SAW
response and the impedance variation of sensors. Some key parameters for
this model are determined by the software FEMSDA, in which the finite element
method (FEM) is used to simulate the mechanical properties of the electrodes
and spectral domain analysis (SDA) is used to model the substrate. SAW devices
have been designed, fabricated and measured. The simulation results agree
well with the experimental results. The applications for capacitive and resistive
sensors are presented.
This work is mainly funded by the German Research Council (DFG—
Graduiertenkolleg Sensorik). The authors are thankful to the research group of
Prof. Ken-ya Hashimoto for their free software FEMSDA.

118
Resonant membrane devices were designed and micro-machined using standard
photolithography and deep silicon etching. The devices were designed to operate
with the high velocity symmetric Lamb mode excited by means of longitudinal
wave (LW) transducers. SAW based designs having varying cavity lengths and
transducer topology were fabricated and characterised. Quality factors well in
excess of 2000 have been demonstrated at a frequency of around 850 MHz.
The Lamb excitation by means of LW transducers demonstrated high efficiency.
These results indicate that LW transducers represent a competitive alternative
to the commonly used interdigital transducers since they offer reduced transducer
size minimizing thus the device area.
This work is supported by the Swedish Foundation for Strategic Research (SSF)
through the ICTEA program in microelectronics.

P1K-3
AMPLITUDE AND PHASE MEASUREMENT OF
SURFACE ACOUSTIC WAVES WITHIN A SAW FILTER
HAVING FAN-SHAPED TRANSDUCERS AND
NUMERICAL SIMULATIONS.
T. CHIBA*, Meisei University, Hino, Tokyo, Japan.
Corresponding e-mail: chiba@[Link]
Several authors have reported optical measurement of surface acoustic waves
within a SAW filter having fan-shaped transducers. P. Dufilie et al [1] and K.
Kokkonen et al [2] investigated the X-112Y LiTaO3 substrate, and H. Yatsuda et
al [4] the Y-Z LiNbO3 and 128Y-X LiNbO3 substrates. Numerical simulations
have also been performed for the field distributions of surface acoustic waves
and electrical frequency characteristics of SAW filters; P. Dufilie et al [1] and K.
Kokkonen et al [2] using the equivalent circuit model, and H. Yatsuda et al [3][4]
using the angular spectrum method. However, the behavior of surface acoustic
waves has still not been clarified in detail, for example, with regard to fan-shaped
transducers and their diffraction effect.
In this paper, we apply a new method of laser heterodyne interferometry to
measuring and processing data for the phase measurement of surface acoustic
waves within a SAW filter having fan-shaped transducers on a 128Y-X LiNbO3
substrate. The following results for the wave motion and the diffraction effect
are clarified. That is, standing waves inside the fan-shaped transducers are
found to move towards the open end of the fan, which is perpendicular to the
propagation direction of the main beam, and to cause several diffraction beams.
Furthermore, a simple numerical simulation concerning the wave motion and
electrical characteristics of a SAW filter having fan-shaped transducers is
proposed. The method uses a parabolic approximation, where the propagation
direction of the wave is corrected by the power flow angle. Numerical simulations
are applied to the field distributions of surface acoustic waves, the wave motion
inside fan-shaped transducers, and the amplitude- and group-delay-frequency
characteristics of the SAW filter. Good agreements are obtained between the
simulations and the experiments.

121
References:
[1] P. Dufilie, [Link], J.V. Knuuttila, [Link] and M.M. Salomaa, IEEE Ultras.
Symp. Proc., pp. 41-44, Oct. 2001. [2] K. Kokkonen, P. Dufilie, J.V. Knuuttila
and M.M. Salomaa, IEEE Ultras. Symp. Proc., pp. 1339-1342, Aug. 2004. [3] H.
Yatsuda, IEEE Ultrasonic Symposium Proc., pp. 173-176, Oct. 2000. [4] H.
Yatsuda, S. Kamiseki and T. Chiba, IEEE Ultras. Symp. Proc., pp. 13-17, Oct.
2001.

P1K-4
DESIGNING CAPACITIVELY COUPLED
MICROELECTROMECHANICAL FILTERS.
A. ALASTALO* and V. KAAJAKARI, VTT Information Technology, Espoo, Finland.
Corresponding e-mail: [Link]@[Link]
Miniature mechanical resonators, fabricated with microelectromechanical-
systems (MEMS) technology, are a potential replacement of off-chip filters (such
as SAW and FBAR) as they are compact in size and integratable with IC
electronics [1]. The demonstrated quality factors of MEMS resonators, Q>100
000 at 10 MHz [2] and Q>1000 at 1 GHz [3], are comparable to their macroscopic
counterparts. While the mechanical properties of microresonators are very
promising, obtaining a low-enough electrical impedance is challenging, especially
for high frequency resonators. To obtain low impedance levels with electrostatic
coupling, a narrow electrode gap (< 200 nm) is needed. This introduces higher
harmonics due to the inverse relationship between the electrode capacitance
and the gap spacing.
In filters, signal intermodulation due to odd-order nonlinearities is especially
detrimental as it can lead to unwanted frequency components within the filter
passband. For example, cubic mixing of two signals having frequencies f1 and
f2 results in third-order intermodulation products at frequencies 2f1-f2 and 2f2-
f1. If f1 = f0 + df and f2 = f0 + 2df, the IM product at 2f1 - f2 is at the passband
center frequency f0 corrupting the desired signal. We have derived analytic
closed-form equations to estimate the third-order intermodulation in MEMS filters
and validated the theoretical results in measurements [4,5].
In this paper, we formulate a design procedure for MEMS filters. Central
characteristics, such as insertion loss, bandwidth and filter distortion, are related
to basic parameters, such as center frequency, mechanical resonator geometry,
supply voltage and receiver architecture, for a single-stage MEMS filter. It is
shown that it is extremely challenging for MEMS filters to operate in the 50 ohm
source and load terminated environments, where FBAR filters are replacing
SAW filters and becoming integrable with CMOS electronics. We show this by
considering the lower limit for the smallest achievable electrode gap spacing
set by intermodulation distortion, electromechanical instability (pull-in) and field-
emission breakdown. The maximum electrode gap, on the other hand, is given
by the insersion-loss specification that set the required mechanical impedance.
These two requirements for the minimum and maximum gap may not be
simultaneously met at GHz frequencies. It is concluded that MEMS filters are
more suited for novel integrated receiver architectures where the impedance

122
levels at both sides of the filter can be much higher than 50 ohm and where the
resonator Q value can be utilized for voltage gain.
[1] C. T.-C. Nguyen, IEEE Trans. Microwave Theory Tech., 47(8), pp. 1486-
1503, Aug. 1999.
[2] V. Kaajakari, T. Mattila, A. Oja, J. Kiihamäki, and H. Seppä, IEEE Electron
Device Lett., 25(4), pp. 173 - 175, Apr. 2004.
[3] J. Wang, Z. Ren, and C. T.-C. Nguyen, IEEE Trans. Ultrason., Ferroelect.,
Freq. Contr., 51(12), pp. 1607 - 1628, Dec. 2004.
[4] A. T. Alastalo and V. Kaajakari, IEEE Electron Device Letters, 26(5), pp. 289
– 291, May 2005.
[5] A. T. Alastalo and V. Kaajakari, IEEE Journal of Microelectromechanical
Systems, submitted.

P1K-5
THIRD-ORDER INTERMODULATION DISTORTION
IN CAPACITIVELY-DRIVEN VHF
MICROMECHANICAL RESONATORS.
Y.-W. LIN*, S.-S. LI, Z. REN, and C. T.-C. NGUYEN, University of Michigan, Ann
Arbor, MI.
Corresponding e-mail: ywlin@[Link]
Substantial increases by more than 25dB in the third-order input intercept points
(IIP 3) of capacitively-transduced MEMS-based vibrating micromechanical
resonators have been attained by using contour-mode disk geometries to replace
previous clamped-clamped beam versions [Navid, et al., MEMS 2001].
Specifically, a 160-MHz contour-mode disk resonator with Q = 20,500 exhibits a
measured IIP3 = 22.11dBm, which is substantially better than the -3dBm
previously measured for a 10-MHz clamped-clamped beam resonator, and which
now erases any lingering skepticism regarding the linearity of micro-scale
mechanical resonators. Indeed, with IIP 3’s greater than 20dBm, high Q
communication filters using the micromechanical resonators of this work should
now be able to replace present-day receive path filters with little degradation
(only 0.11dB) in cumulative receiver IIP3.
The 160-MHz micromechanical contour-mode disk resonator measured in this
work was fabricated via a fully planar wafer-level surface-micromachining process
and consists of a 3µm-thick, 17µm-radius polysilicon disk suspended by a stem
at its center, and enclosed by two polysilicon capacitive transducer electrodes
spaced 70nm from the disk perimeter. To operate the device, a dc-bias voltage
is applied across its electrode-to-resonator capacitive gaps, and an ac excitation
applied to the electrodes. Together, these voltages drive the disk into a vibration
mode shape where it expands and contracts uniformly around its perimeter.
The disk motion creates dc-biased time-varying electrode-to-resonator
capacitors, which then source output currents that are linear for small motions,
but that can contain distortion components when vibration amplitudes are large
enough to instigate capacitive transducer nonlinearity. In particular, when two
electrical interferer tones at fo-200kHz and fo-400kHz are applied to two separate

123
input electrodes, these signals and their induced mechanical displacements
can interact to generate third-order intermodulation forces within the resonator
passband, that then generate third-order intermodulation output distortion
components, and thus, lower the IIP3.
The IIP3 = 22.11dBm measured using the above “two tone” test for a 160-MHz
contour-mode disk with Q = 20,500 under a dc-bias of 7V is quite close to the
value of 19.07dBm predicted by a model derived via harmonic balance analysis
to be expanded upon in the full paper. Among the IIP3 dependencies correctly
predicted by this model are: (1) an increase in power IIP3 with increasing resonator
dc-bias voltage, verified by measured IIP3’s of 19.6dBm, 21.9dBm, and 24.7dBm
for a 100-MHz contour-mode disk resonator under dc-biases of 3V, 5V, and 7V,
respectively; (2) an increase in power IIP3 with decreasing electrode-to-resonator
gap spacing; and (3) an increase in power IIP3 with increasing resonator size,
verified by measured of IIP3’s of 25.0dBm, 21.9dBm, and 20.1dBm for 60-MHz,
100-MHz, and 160-MHz contour-mode disk resonators with radii of 44µm, 26µm,
and 17µm, respectively. The model further predicts clear paths towards even
better IIP3’s, perhaps up to 20dB better.

P1K-6
CHARGE-BIASED VIBRATING
MICROMECHANICAL RESONATORS.
S.-S. LI*, Y.-W. LIN, Y. XIE, Z. REN, and C. T.-C. NGUYEN, University of Michigan,
Ann Arbor, Michigan.
Corresponding e-mail: ssli@[Link]
MEMS-based capacitively-transduced micromechanical disk resonators have
been demonstrated that dispense with conventional high voltage source direct
dc-biasing and instead use stored charge on their resonant structures to achieve
equivalent performance. In particular, a 100-MHz micromechanical contour-mode
disk resonator has been demonstrated with the same Q (~25,000) and series
motional resistance when charge-biased, as when biased via a dc voltage source
directly connected to its resonant structure. In removing the need for a dc voltage
source, charge-biasing affords several important advantages, including: (1) better
immunity against power supply variations, which should greatly facilitate
integration into mixed-signal analog-digital communication circuits; (2) an ability
to supply an effective bias voltage higher than the system power supply without
the need for power hungry charge pumping; and (3) removal of the interconnects
that would normally be needed to supply the dc-bias voltage to the resonant
structure. These advantages stand to greatly simplify the use of micromechanical
resonator technology in portable systems.
The micromechanical disk resonator used in this work to compare charge- and
direct dc-biased performance comprise 26µm-diameter doped-polysilicon disks
suspended 650nm from the substrate by stems at their centers, with two doped-
polysilicon electrodes spaced 80nm from the disk edges, each covering half of
the disk’s circumference. Under conventional operation, a direct dc-bias voltage
is applied to the suspended structure through an electrode connected to its
conductive stem, and an ac excitation vi is applied to one electrode. The two
voltages combine to generate a force at the frequency of vi that drives the disk

124
into a resonance mode shape where it expands and contracts, generating a
time-varying electrode-to-resonator capacitance. An output current then ensues
that traces out a very high Q bandpass biquad transfer function as the input
frequency is swept.
In this work, the lead from the resonant structure to the dc-bias voltage source
is abruptly broken, leaving charge on the resonator structure to maintain its bias
state. Since the resonator is electrically well isolated from its surroundings, this
charge leaks out very slowly, to the point where the resonator is still operational
for long periods. The operational period can be made even longer by tying a
capacitor in parallel with the resonator structure, which can increase the amount
of charge available by several orders of magnitude, hence increase the
operational lifetime by the same order. For a stand-alone charge-biased
resonator, leaky discharging of the resonant structure in vacuum attenuates the
output signal by 3dB after 900s. If a 0.18µF capacitor is charged and attached
to the floating resonant structure, the resonator can operate for more than 8.5
days before its signal is attenuated by 3dB, which clearly illustrates the long-
term efficacy of this charge-biasing approach. Frequency shifts caused by
discharge-induced bias shifts are less than 2.5 ppm per hour, making plausible
a “charge-and-refresh” operating mode with one-hour refresh intervals.

P1K-7
LAYER MODE DEVICES ON EPITAXIALLY GROWN
GAN FILMS ON AL2O3 .
K. HOHKAWA* 1 , C. KANESHIRO 1 , K. KOH 1 , K. NISHIMURA 2 , and N.
SHIGEKAWA2, 1Kanagawa Institute of Technology, Kanagawa , Japan, 2Nippon
Telegraph & Telephone Co., Kanagawa, Japan.
Corresponding e-mail: chinami@[Link]
The layer modes on thin film, such as Sezawa mode, have an advantage over
the Rayleigh mode, because they provide higher frequency devices using the
same inter-digital transducers. However, these modes have the velocity
dispersion, which might deteriorates device characteristics, especially on wide
bandwidth devices. Device characteristics also tend to be dispersed by many
factors on fabrication processes. This paper investigates a feasibility of layer
mode devices with a high frequency and a wide bandwidth using present process
technologies available for us, and tries to investigate a simple design method
for these devices.
We have fabricated a simple delay line on some kinds of GaN films (including a
commercially available film) and measured device characteristics of layer modes
around a few GHz frequency range. We have adopted the film thickness of
shorter than the periodic length of IDT (less than 1/1.6). In this condition the
layer modes are seriously affected by an isotropy of Al2O3 substrate, we selected
propagation directions along the main axis of sapphire and its perpendicular
direction. We have measured device characteristics focused on amplitude and
delay. The results have shown that the device characteristics are affected
seriously by the difference on the propagation direction and residual stresses in
film. We obtained many interesting results, such as the propagation wave along

125
main axis has shown a negative dispersion characteristics (the delay time
decreases when operation frequency increases).
Based on these results, we clarify a condition suitable for device application.
We also propose a simple design method for layer mode device, based on the
impulse response model and discuss its feasibility.

P1K-8
DEVELOPMENT OF HIGH-SPEED LASER PROBE
SYSTEM BASED ON KNIFE-EDGE METHOD FOR
DIAGNOSIS OF RF SAW DEVICES.
H. KAMIZUMA*, T. OMORI, K. HASHIMOTO, and M. YAMAGUCHI, Chiba
University, Chiba-shi, Japan.
Corresponding e-mail: [Link]@[Link]
This paper describes the development of a high-speed laser probe system for
diagnosis of RF surface acoustic wave (SAW) devices. A fast scanning rate of
2.5 kS/s is realised, making it possible to visualise an SAW field consisting of
2,620×410 data points in approximately 40 min.
For the optical detection, the phase-sensitive knife-edge method was employed.
This makes the system very insensitive to low-frequency mechanical vibrations
caused by the fast stage translation. A prism mirror was used as a knife-edge
and combined with a high-speed photo diode (PD) for system operation in GHz
range.
A heterodyne receiver system was newly developed for the synchronous
detection of the PD output and the high-speed data acquisition described below.
For the fast measurement, the following scanning technique was applied: the
mechanical stage is translated continuously instead of the conventional “inch
worm” action, and the detector output is acquired successively by a high-speed
data-logger. Trigger pulses are generated from the output of a high-precision
linear-scale installed in the translation stage and fed to the data-logger to
synchronise with the stage movement. Then the translation stage can be driven
in its maximum speed of 1 mm/s without degradation of the measured result.
After describing the system setup in details, the system is applied for the
characterisation of spurious resonance modes in SAW resonators on ST-cut
quartz and 15oYX-LiNbO3. In conjunction with skillful use of image processing in
wavenumber domain, it is shown how the present system is effective in the
diagnosis and development of SAW devices.

126
P1K-9
NONSTANDARD ELECTROSTATIC PROBLEM FOR
SAW INTERDIGITAL TRANSDUCER IN EXTERNAL
ELECTRIC FIELD.
Y. TASINKEVYCH* and E. DANICKI, Institute of Fundamental Technological
Research, Polish Academy of Sciences, Warsaw, Poland.
Corresponding e-mail: yurijtas@[Link]
The nonstandard electrostatic problem for the system of strips making IDT placed
in external electric field is solved by a novel method. It is based on earlier
presented approach [1] exploiting the special template functions constructed in
spectral domain and satisfying the electric boundary conditions on the strips.
The functions can be effectively evaluated numerically with sufficient accuracy
[2]. What is most important, the same functions and analogous approach
exploiting them can be applied in the analysis of the electrostatic problems for
anisotropic media, and also numerous problems of electromagnetic and elastic
wave scattering by strips or planar cracks.
In contrast to standard electrostatics, here we construct solutions in the spectral
domain. Note that the field spatial distribution is the least important in applications:
measured are either the mutual strip capacitances depending on the total strip
charges (the integrals of the charge distribution), or the spatial spectrum (in the
scattering cases, like Bragg scattering by strips or frequency characteristics of
surface wave transducers). The solution constructed in spectral domain with
satisfactory accuracy is advantageous as compared with the standard evaluation
of spatial charge distribution at least in the latter case, because of high sensitivity
of the charge spectrum on the charge distribution details over the entire system.
Having the spatial spectrum, one can easily evaluate the spatial distribution
and its integral (in order to evaluate the measured total charges of strips) with
similar high accuracy, what is not true in the reverse way. Moreover, the inverse
Fourier transformation yields the tool for verification of the spectral results, lacking
in the standard analysis.
1] E. Danicki, “Strip electrostatics - spectral approach”, 1996 IEEE Ultras. Symp.
Proc. (1996) 193-196.
[2] [Link], “Numerical efficiency of interdigital transducers charge spatial
spectrum evaluation methods”, PhD Thesis, Polish Acad. of Sci., IFTR, Warsaw,
2004.
Present work was sponsored by State Committee of Scientific Research, Grant
3 T11B 046 26.

127
Session: P1L
MATERIALS FOR TRANSDUCERS
Chair: T. Shrout
The Pennsylvania State University
P1L-1
TEMPERATURE EVALUATION OF SOFT AND HARD
PZT TRANSDUCERS FOR ULTRASONIC TRAPPING IN
A MICROFLUIDIC PLATFORM.
L. JOHANSSON* 1, M. NILSSON 2, T . LILLIEHORN 1, M. ALMQVIST 2, J.
NILSSON2, T. LAURELL2, and S. JOHANSSON1, 1Uppsala University, Uppsala,
Sweden, 2Lund University, Lund, Sweden.
Corresponding e-mail: [Link]@[Link]
This paper reports a comparison of soft and hard piezoceramic transducer
materials used for ultrasonic standing wave particle trapping in a microfluidal
bioanalytical platform. The investigation is made with the objective to obtain
high acoustic forces with a minimum of temperature increase. Temperature is a
critical parameter for bioassays and most often needed to be kept below a certain
level to allow handling of cells and proteins. The main conclusions in this paper
are that it is possible to get efficient trapping with a temperature increase of only
a few degrees and that a soft piezoceramic material has advantages in an
application such as this.
Several groups have reported using acoustic forces in fluidic systems for
separation or trapping of particles and cells in macro-scale resonators. In our
group, a micro-scale system has been developed using higher frequencies and
thus stronger trapping force. The platform with an array of individually controlled
trapping sites offers a versatile system for various bioassays.
Miniature PZT-ultrasound transducers are integrated in the bottom of a
microchannel and are designed to work as acoustic resonators at a frequency
of around 12 MHz. Properly matched channel height allows a standing wave to
be formed between each transducer and a glass-reflector with a local pressure
minimum in the center of the channel defining the trapping site. Temperature
rise in the channel is caused by mechanical and dielectric loss in the PZT
transducers. At resonance and at the conditions suitable for efficient trapping,
the standard measurements of dielectric loss tanδ and mechanical loss Q-1 are
insufficient for predictions of what material is the most favourable with respect
to both temperature and trapping force in our device. To be able to select the
optimal transducer material, the acoustic output and the fluid temperature has
to be measured.
The piezoelectric materials (EDO EC-69 and EC-76) have different piezoelectric
constants and the acoustic pressure is therefore measured, with a miniature
hydrophone and at a distance of 35 wavelengths from the transducer, and the
drive voltage adjusted to give the same output. For these drive voltages, fluid
temperature in channel is evaluated by measuring fluorescence intensity of
temperature sensitive Rhodamine B dissolved in the fluid.

128
The piezoelectrically softer material delivers more than twice as high output
pressure for a drive voltage of 7 V. For both materials, series resonance gives
about 10 % higher pressure than parallel resonance. Taking temperature also
into account, the soft material gives a lower temperature increase (2 ° C) than
the hard material (7 ° C). At anti-resonance however, the hard material is more
favourable (plus 3 ° C) than the soft material (plus 5 ° C). So most favourable is
to run the soft material at resonance frequency followed by hard material at
anti-resonance frequency. The loss mechanisms in these transducer materials
are complex and the paper presents some of the issues that have to be
considered in these ultrasonic trapping devices.

P1L-2
MULTILAYER PIEZOELECTRIC
COPOLYMER TRANSDUCERS.
T. LILLIEHORN*1, T. BLOM1, M. NILSSON2, M. ALMQVIST2, U. SIMU1, and S.
JOHANSSON1, 1Uppsala University, Uppsala, Sweden, 2Lund Institute of
Technology, Lund, Sweden.
Corresponding e-mail: [Link]@[Link]
This contribution, for the first time at an international conference, presents
process technology to fabricate multilayer ultrasonic transducers of the
piezoelectric copolymer poly(vinylidene fluoride-trifluoroethylene) (P(VDF-TrFE))
by spin-coating. Multilayer technology is used to be able to subject the polymer
material to high electrical fields at low drive voltages, and to increase the
capacitance of the transducers for impedance matching purposes. Multilayer
transducers become increasingly important in miniaturized systems, due to the
low permittivity of the piezoelectric polymer material. The outlook of the present
work is to investigate if miniaturized polymer transducers can be used to produce
continuous wave acoustic energy high enough to be able to manipulate particles
in a microfluidic system by acoustic radiation forces in an acoustic standing
wave.
Multilayer transducers, 6 mm in diameter, have been fabricated on electroded
glass substrates by spin-coating of a solution of P(VDF-TrFE) with a molar ratio
of 65:35 and methyl ethyl ketone (MEK). Each spin-coated layer was 17 µm
thick, and each transducer consisted of three polymer layers. After each layer,
the substrate was annealed at 120°C for three hours, followed by the deposition
of an evaporated electrode layer. The annealing of each layer was critical to
being able to deposit another layer on top without dissolving the underlying
polymer material and destroying the intermediate electrode layer. Single layer
transducers were fabricated by a similar process scheme, but with electrodes
deposited only on the substrate and on top of the third spun polymer layer. The
transducers were polarized at peak electric fields between 19-76 V/µm, to analyze
the effect of varying polarization.
The electrical impedance spectra of the transducers clearly showed the
resonances in the glass substrate. The center frequency of the transducers was
found to be 10.4 MHz, and the impedance of the multilayer transducer at this
frequency was 50 Ω, compared to 400 Ω for the single layer transducer, being
better matched to 50 Ω electronics. The transducers were characterized by

129
pitch-catch measurements, using polymer transducers as both transmitters and
receivers. The round-trip insertion loss (IL=20log(VD/VP)) was calculated from
the peak-peak voltage of the drive signal (VD), and the peak value of the pitch-
catch response (VP). The results were corrected for the gain in the receiving
electronics and for the dampening in the medium. The IL decreased with 12.5
dB when increasing the poling field of the multilayer transducers from 19 V/µm
to 76 V/µm. At 50 V/µm polarization, the IL was 11.7 dB less for the multilayer
transducers when compared to the single layer transducers.
To conclude, multilayer polymer transducers have successfully been fabricated
by a process technology based on spin-coating. The multilayer transducers have
been evaluated by pitch-catch measurements, and have shown to outperform
similar single-layer transducers.

P1L-3
MODELLING OF ULTRASONIC WAVE PROPAGATION
IN INTEGRATED PIEZOELECTRIC STRUCTURES
UNDER RESIDUAL STRESS.
M. LEMATRE*, G. FEUILLARD, and M. LETHIECQ, LUSSI GIP Ultrasons
Université F. Rabelais, EIVL Blois France.
Corresponding e-mail: gfeuillard@[Link]
The development of piezoelectric integrated structures is mostly based on thin
or thick film technology. In this case, the piezoelectric layer is laid down on a
substrate and submitted to thermal treatment. In particular, after the sintering
process, due to the difference of thermal expansion coefficients between the
film and the substrate, a non uniform shrinkage appears that leads to a
degradation of the functional properties of the structure through residual stress.
Thus, it has become necessary to understand the role of residual stress in
piezoelectric layers in order to predict the performance of integrated structures.
In a first part, a bulk piezoelectric material with internal stress is considered. For
this, Christoffel’s equation for a piezoelectric material is modified to take into
account a uniform residual stress on a given cross-section. A numerical study of
its influence is led on the slowness curves and coupling coefficients of a PZT-
based material. It is showed that a lateral residual stress could couple pure
transverse mode to piezoelectricity because of a slight rotation of the polarisation
vector. In a second part, modified Cristoffel tensor is used to calculate the
dispersion curves of Lamb waves in a piezoelectric plate. The Lamb modes are
found to be sensitive to the residual stress. In particular, we show how the
behaviour of the first Lamb modes is modified with residual stress.
In a third part, these results have been extended to a piezoelectric film laid
down on a substrate in order to model the importance of these phenomena on
the behaviour of an integrated structure.
This work was supported by the European community: MINUET Project NMP2-
CT-2004-505657

130
P1L-4
NET-SHAPE CERAMIC MANUFACTURING AS AN AID
TO REALIZE ULTRASONIC TRANSDUCERS FOR HIGH
RESOLUTION MEDICAL IMAGING.
T. BUTTON*1, S. COCHRAN2, K. KIRK2, K. MCDONALD2, C. MEGGS1, D.
RODRIGUEZ-SANMARTIN1, R. WEBSTER1, and D. ZHANG1, 1University of
Birmingham, Birmingham, UK, 2University of Paisley, Paisley, Scotland, UK.
Corresponding e-mail: coch-ph0@[Link]
High frequency ultrasonic transducers are in demand for medical imaging
procedures requiring high spatial resolution. However, cost-effective fabrication
for frequencies above approximately 20 MHz is challenging because of the need
for a thin layer of active material. For example, PZT 5A ceramic must be
approximately 100 µm thick and a typical piezoceramic-polymer composite
approximately 80 µm thick to operate at 20 MHz. These thicknesses are difficult
because they exceed the capabilities of typical thick and thin film processes but
are sufficiently thin that lapping bulk materials is expensive and inefficient. In
this paper, the alternative to use net shape ceramic processing is reported. For
thin piezoceramic material, the most conceptually simple net-shape processing
route is to use viscous polymer processing (VPP) to produce green-state material
with rheological properties compatible with calendaring to produce thin sheets
which are dried, sintered and can then be cut to shape mechanically or using
more sophisticated techniques such as laser machining. This process is also
potentially compatible with mechanical shaping, for example to produce a
focusing element. The application of electrodes within the net-shape process is
also possible, or thin film electrodes can be deposited afterwards. Although
such thin specimens are difficult to handle, the immediate addition of a supportive,
acoustically-appropriate backing material to lend strength, allows the use of
straightforward processes to configure the device into the final ultrasonic
transducer. In the present paper, we report piezoceramic made with TRS600FG
material finished to thicknesses of 50 µm and 110 µm. The behaviour of the thin
samples has been found to be similar to bulk material made with TRS600FG, for
example with a thickness mode coupling coefficient, kT, of 0.52 and a relative
permittivity, εRS, of 1540. This material has been used as the basis of prototype
ultrasonic transducers with element diameters of a few millimetres, corresponding
to an electrical impedance in the range 50 - 70 Ω when operated at frequencies
approaching 50 MHz. Testing has been performed underwater with a JSR
DPR300 Pulser-Receiver, utilising a broadband membrane hydrophone and in
pitch-catch mode with two transducers. The successful results suggest that net-
shape ceramic manufacturing can overcome the difficulties presently associated
with the fabrication of thin piezoceramic.
The authors acknowledge the contribution of [Link] to the work reported
here.

131
P1L-5
BROADBAND EMFI ULTRASONIC TRANSDUCER FOR
BAT RESEARCH.
A. STREICHER*1, M. KALTENBACHER1, H. PEREMANS2, and R. LERCH1,
1
Friedrich Alexander University Erlangen-Nuremberg, Erlangen, Germany,
2
Antwerp-University Faculty St. Ignatius, Antwerp, Belgium.
Corresponding e-mail: [Link]@[Link]
In the field of bat research and robot navigation, broadband air ultrasonic
transmitters and receivers are of great importance. For example, in order to
investigate the sound as produced by a wide variety of bat species, the
transducers have to operate with a frequency range of 20 - 200 kHz and have to
match sensitivity (receiver) as well as transmit efficiency (transmitter) of living
bats. These requirements can not be achieved by commercially available in-air
ultrasonic transducers due to their small bandwidth. Hence, a new transducer
technology is required to build up broadband ultrasonic transducers with a good
adaptation to air. The most promising transducer material for this is a cellular
polymer film called electro mechanical film (EMFi). The cellular structure of the
material results in a relatively high piezoelectric constant d33 up to 800 pC/N
and a resonance-frequency of about 300 kHz.
By utilizing this polymer material different transmitters with a diameter of 1.5 cm
were developed for emitting a chirp signal with a sound pressure level up to 90
dB at a distance of 1m for the whole frequency range of 20 - 200 kHz. With the
same material, a broadband ultrasonic receiver with a sensitivity of 500 µV/Pa
and a low equivalent acoustic noise level of 45 dB was set up. For an optimisation
of the transmitter and the receiver we need a deeper understanding of the
physical behaviour of the polymer material. Therefore, we applied a 3D finite
element simulation by using a piezoelectric material model for a macroscopic
description of the EMFi material. However, vibration measurements of the
transducer surface show a nonlinear inhomogeneous vibration behaviour at
and above resonance frequency. One reason for this is the inhomogeneous
structure of the foil. Inside the polymer film the number of cavities as well as
their size strongly varies. Because the resonance frequency of each point of the
surface depends on the average cavity size at this point, the whole surface
vibrates inhomogeneous. Therefore the EMFi material can not be described
with a piezoelectric material model and we developed a more complex
microscopic model for the precise numerical simulation. To solve this problem
we computed the electrostatic and mechanical partial differential equation
coupled by the electrostatic forces (Coulomb forces) including the complex
geometric structure of the cellular ferroelectric film. To investigate the influence
of the geometric structure on the vibration behaviour, models with different void
shapes and sizes have been taken into account. For practical simulation of the
whole transducer behaviour the microscopic model is not applicable due to
computer resources. Therefore, we currently investigate in the development of
a macroscopic model for this purpose.
In our paper, we will discuss the simulation results for various designs of
transmitters and receivers as well as corresponding measurement results.

132
P1L-6
BASIC STUDY ON LEAD FREE BNT PIEZOELECTRIC
FILM DEPOSITED BY HYDROTHERMAL METHOD.
T. HASEGAWA*1, N. KAWASHIMA1, S. TAKAUCHI1, M. ISHIKAWA2, and M.
KUROSAWA2, 1Toin University of Yokohama, Yokohama, Kanagawa, Japan,
2
Tokyo Institutu of Technology, Yokohama, Kanagawa, Japan.
Corresponding e-mail: shin1@[Link]
PZT piezoelectric poly-crystalline film synthesized by hydrothermal method has
been studied actively in order to develop ultra-miniature actuators or sensors in
our laboratory. This PZT film has some advantages that it shows piezoelectricity
without a polling process, it can be deposited on the titanium substrates with
complex shape. However, it is apprehended recently that PZT has malign
influences on life and environment, because PZT is chemical compound including
lead (Pb). Therefore, we tried to deposit lead free BNT [(Bi0.5, Na0.5) TiO3]
piezoelectric film on titanium substrates by hydrothermal method. Bismuth oxide,
sodium nitrate and potassium hydroxide were used as source material for lead
free BNT piezoelectric film in this study. It was confirmed by SEM image, XRD
(X-ray diffraction) pattern and EDS (Energy dispersive X-ray spectrometer) data
that the deposited films on the titanium substrate were poly-crystals with perovskit
structure. However, the synthesized poly-crystalline film had extremely
unbalanced composition rate of Bi 10. We fabricated bimorph vibrators by
depositing lead free BNT film on titanium substrate by hydrothermal method. It
was found by measurements with laser Doppler vibrometer (LDV) that vibration
velocities of the bimorph vibrator with BNT film were about from one fifth to one
tenth of vibration velocity of bimorph vibrator with PZT film deposited by
hydrothermal method with same size. These results were reported at 2004 IEEE
UFFC Joint 50th anniversary conference in Montreal. We repeated the
experiments with adjusting Bi concentration in the source material solution after
the 2004 conference in Montreal. It was found as a result that composition of
synthesized poly-crystalline film could be improved to Bi 20 at the condition of
Bi concentration of 0.1 M. It was proved by LDV measurement that a bimorph
vibrator using the poly-crystalline film with composition of Bi 20 showed the
vibrating velocity of twice the conventional data. We will reports these new
considered results in this symposium.

P1L-7
PIEZOELECTRIC PROPERTIES OF BATIO3 THIN FILMS
GROWN BY ECR-PLD.
S. ITO, K. NAKAMURA*, and K. ISHIKAWA, Tohoku University, Sendai, Japan.
Corresponding e-mail: nakamura@[Link]
Ferroelectric (001) BaTiO3 thin films are suitable for generation of high-frequency
longitudinal waves, because of the high electromechanical coupling factor kt of
about 0.61 and moderately low dielectric constant of about 56. Epitaxial growth
of BaTiO3 thin films has been reported in a number of papers so far. However,

133
there has been no report on their piezoelectric properties related with the high-
frequency thickness-extensional mode. We have recently succeeded in epitaxial
growth of high quality (001) BaTiO3 films on conductive La-doped (100) SrTiO3
substrates using an electron-cyclotron-resonance-assisted pulsed laser
deposition (ECR-PLD) technique. This paper reports the growth procedure and
the piezoelectric properties related with high-frequency thickness-extensional
vibrations. The X-ray diffraction analysis revealed that the lattice constant
perpendicular to the surface was very close to the c-axis lattice constant of bulk
crystals. The FWHM of the rocking curve of a (001) BaTiO3 film was as small as
0.09°, which is the smallest among those reported up to now. The electrical
admittance response for the thickness-extensional modes in a high-frequency
range was measured for the first time. In the 1-8 GHz range, there were observed
many large resonance peaks associated with the overtone thickness-extensional
vibrations of the whole BaTiO 3-film/substrate composite structure. The
electromechanical coupling factor kt of the BaTiO3 film under a 20V dc-bias was
evaluated to be 0.41 by comparison with equivalent circuit simulations. The
dependence of measured ferroelectric properties on the dc-bias voltage is also
presented.

P1L-8
ALUMINUM NITRIDE THIN FILMS ON TITANIUM FOR
PIEZOELECTRIC MEMS APPLICATIONS.
S. BOESHORE*, E. PARKER, V. LUGHI, and N. MACDONALD, University of
California, Santa Barbara, CA,.
Corresponding e-mail: sethb@[Link]
Highly-textured aluminum nitride thin films have been deposited onto titanium
substrates for the purpose of fabricating piezoelectric MEMS. As a demonstration
of this technology, which could lead to radio-frequency devices suitable for harsh
environments and broadband high-g accelerometers, trilayer cantilever beams
(AlN films sandwiched between titanium bottom electrodes and aluminum upper
electrodes) were fabricated using standard MEMS bulk micromachining
techniques. We present here the first known demonstration of a piezoelectric
titanium MEMS.
Piezoelectric aluminum nitride films were deposited onto bulk polycrystalline
titanium substrates using a high-power, low-temperature 40-kHz AC sputter
deposition. Because sputtered thin films are polycrystalline the piezoelectric
coefficient is controlled by the film texture; in this case, the texture is the alignment
of (002) grains normal to the substrate. The deposition parameters were
optimized to produce the maximum (002) grain alignment in 1.5- to 2-µm-thick
AlN films. X-ray rocking curve analysis of the deposited AlN films showed that
full-width half-maxima (FWHM) of under 4° were readily achievable and indicated
that the films were suitable for fabricating piezoelectric devices.
To fabricate the trilayer beams, the AlN thin film was etched with a chlorine-
based plasma etch followed by a re-entrant deep etch into the titanium substrate
to a depth of approximately 10 µm. The beams were then released by an isotropic
wet etch in dilute hydrofluoric acid, and finally the top contact layer of 300 nm of
aluminum was deposited by electron beam deposition. The released beams
measured 60 µm long by 10 µm wide, and had a total maximum thickness of
134
approximately 4 µm.
Testing of the released trilayer beams was performed using a laser vibrometer
to directly measure beam velocities caused by an applied voltage. Because of
the piezoelectric coefficient d31 , which couples a z-axis electric field to an x-y
strain, a voltage across the AlN film created a strain, resulting in beam
displacement. An oscilloscope was used to integrate the velocity data into beam
displacement. Signal output from the vibrometer was also sent into a vector
signal analyzer to record the frequency response of the beams.
The maximum beam tip resonant displacement measured was 3 µm. The
resonant frequency of the beams varied from 800 kHz to over 1 MHz. This
difference was attributed to processing variations, as all of the cantilevers had
the same mask dimensions. The quality factors of the beams were measured at
atmospheric pressure and were in the range of 150 to 350. A finite element
model of the cantilever was created to compare theoretical results to the
measured displacement data. Using an extrapolation method, an AlN film with
FWHM of 6.9° was found to have a piezoelectric coefficient d31 of 0.80 C/m2 ,
roughly half of the reported single-crystal value.

Session: P1M
TRANSDUCER MODELING
Chair: R. Lerch
University of Erlangen
P1M-1
ANALYSIS OF POINT SPREAD FUNCTION IN PHASED
ARRAY IMAGING.
MAHMOUD SAKHAEI*, ALI MAHLOOJIFAR, and M. HASAN GHASEMIAN,
Tarbiat Modarres uiversity, Tehran, Iran.
Corresponding e-mail: sakhaei@[Link]
Ultrasound imaging system is generally considered as a linear system and so,
it may be mathematically described by impulse response function (IRF) or, in
image domain, by point spread function (PSF). It is the subject of many articles
to model the ultrasound field in an ultrasound system, but they do not give any
analytic expression of IRF or PSF for a wideband ultrasound array. However
that expression has been given for single transducer and focal or far field region
of an array. So, there is a lack of closed form expression for PSF of an array
which is applicable for any excitation and for all regions.
In this paper, we represent an approximate closed form expression of 2-way, 2-
dimentional wideband PSF for phased array configuration, which may be used
to gain insight to image formation. In addition it is helpful to design the weightings,
pulse shape, bandwidth and frequency and so on. Also it may be used for
modeling signal statistics and determining system characteristics such as
sensitivity, resolution and coherence as well as, image restoration.

135
The model is established on summation of spherical wave produced by each
element and beam spreading and element directivity effects are considered by
a convenient scaling factor. It is demonstrated that our model gives acceptable
results by comparing it to results obtained by Ultrasound Field Simulation
program, FIELDII.
Our model considers the property of spatial shift variant of PSF and is represented
for fixed focus in transmit and receive, however it is applicable to any focusing
configuration such as dynamic focusing and synthetic aperture techniques. It is
known that in each temporal or spatial point, the PSF is a function of focusing
point and the point at where the reflector lies. We demonstrate that there is an
interesting relation between these points which is that if the places of reflector
and focusing point are interchanged, then the new system response is time
reversed of primary response.
We show that 2-way PSF is equal to the convolution of one way PSF’s inverse
filtered by the excitation pulse. Also by analyzing in far field case, it is concluded
that our model is consistent to other forms previously proposed and the PSF is
shift invariant. Moreover, the effective aperture is convolution of transmit and
receive aperture only in far field case.

P1M-2
CONNECTION BETWEEN X-WAVES, FOURIER-
BESSEL SERIES AND OPTIMAL MODELLING
APERTURE FOR CIRCULAR SYMMETRIC ARRAYS.
P. D. FOX*, INSERM U619 & CIT, CHU Bretonneau, 2 Bis Boulevard Tonnelle,
37044, Tours, France., Tours, Centre, France.
Corresponding e-mail: fox@[Link]
This paper addresses various unresolved issues raised in publications [1], [2],
and [3], in connection with the study and application of limited-diffraction and
non-diffracting beams. Nondiffracting beams have the property of a constant
radial profile with propagation distance, subject to an infinite aperture source,
and the related theories advanced in the context of medical imaging have resulted
in the possibility of extremely high fame rates. However, all the fundamental
theory assumes an infinite-aperture source being available. In reality this is not
possible, and when nondiffracting beams are implemented on finite source
apertures they become limited diffraction beams. [1], [2], [3] have studied the
use of and numerical differences between nondiffracting theory and limited
diffraction implementation of such beams for field computation and tuning. In
[1], [2] it was shown that an iterative technique could be applied to extend the
effective modelling aperture from the physical limits of the transducer (limited
diffraction basis functions) towards an infinite modelling aperture (Bessel beams
/ X-waves), to both tune and compute the emitted field from any given circular
symmetric flat array with linear propagation conditions. This technique involved
the concept of a modelling aperture spanning the gap between the physical
limit of the transducer and infinity, which, when increased iteratively resulted in
convergence of the corresponding computed field and source driving function.
However, the technique relied on a combination of iterative combinations and

136
numerical field convergence to within a pre-selected limit in order to terminate
the computations at an appropriate point.
In this paper, a formal mathematical connection between the limited-aperture
(limited diffraction) basis functions and the nondiffracting infinite aperture theory
(Bessel beams / X-Waves) is established as a function of the increasing modelling
aperture. The result is that a specific optimal modelling aperture may then be
specified as a function of frequency spectrum, spatial field extent to be
investigated, and pulse repetition frequency. Consequently the previous iterative
technique may be replaced by a single one-shot computation to achieve the
same result. As a result, the new technique is significantly more efficient than
the previous technique and the specific saving in computation depends on the
particular transducer considered, but typically computational reductions are in
the order of 50%. The global contribution of the paper is twofold : firstly a formal
mathematical connection between limited diffraction beams and nondiffracting
beams as function of increasing modelling aperture, and secondly the derivation
of the optimal modelling aperture required for computation and tuning of circular
symmetric fields with minimal computational demands.
[1] JASA ,Vol. 113, no. 5, pp. 2412-2423, May, 2003.
[2] IEEE Trans. UFFC, Vol. 49, no. 9, pp 1179-1190, Sep 2002.
[3] IEEE Trans. UFFC, Vol. 49, no. 1, pp 85-93, Jan 2002.
This work was partially sponsored by Le Studium, Orleans, France.

P1M-3
FOCUSED AIR TRANSDUCERS: SOUND FIELD
COMPUTATION AND FE MODELING.
J. LAN*, S. FRAZIER, and M. CHIASSON, Airmar Technology Corporation,
Milford, NH.
Corresponding e-mail: jlan@[Link]
This paper describes the design of a transducer for air-borne ultrasonic
applications. The transducer is highly focused with an F-number approaching
one. The application for this air transducer is for non-contact non-destructive
testing where a very small focal spot is required. The transducer uses a
piezoelectric ceramic as the active element. Key design criteria are to maximize
ultrasonic pressure at the focus, optimize depth of field, and be capable of
operating over a reasonable bandwidth. Essential to understanding the tradeoffs
is FE analysis by which displacements along a curved surface can be accurately
predicted (both amplitude and phase). The front of the transducer is composed
of syntactic foam with the desired geometric shape, bonded to a flat disk
piezoelectric ceramic. Syntactic foam is often used as a quarter-wave matching
layer for air transducers. In this application, however, thickness of the syntactic
foam ranges from a fraction of a wavelength near the center to tens of
wavelengths near the outer edge. Geometry of the emitting surface has the
appearance of the interior of a hemisphere. FEA revealed distribution of
displacements over the hemisphere, as a function of frequency around the
transducer’s resonance frequency.

137
In general a target is working in near sound field and amplitude on radiation
surface are not uniform, and are complex number, therefore their sound field
are complicated. A group of formulars for calculating sound field distribution
and sound pressure along symmetrical axis of surface are derived based on
known amplitude on the spherical surface.
Engineering evaluation of the transducer for NDT has demonstrated that good
resolution for defects near the surface (which is the system objective).

P1M-4
DIFFERENT APPROACHES TO FINITE ELEMENT
MODELING OF EFFECTIVE MODULI OF POROUS
PIEZOCERAMICS WITH 3-3 (3-0) CONNECTIVITY.
A. NASEDKIN1, A. RYBJANETS*2, L. KUSHKULEY2, Y. ESHEL2, and R. Tasker3,
1
Rostov State University, Rostov on Don, Russia, 2UltraShape Ltd., Tel Aviv,
Israel, 3TASI Technical Software Inc., Kingston, Canada.
Corresponding e-mail: andrey58@[Link]
Theoretical aspects of the effective moduli method for an inhomogeneous
piezoelectric media were examined. Four static piezoelectric problems for a
representative volume that allow finding the effective moduli of an
inhomogeneous body were specified. These problems differ by the boundary
conditions which were set on a representative volume surfaces: a) mechanical
displacements and electric potential, b) mechanical displacements and normal
component of electric flux density vector, c) mechanical stress vector and electric
potential, and d) mechanical stress vector and normal component of electric
flux density vector. Respective equations for calculation of effective moduli of
piezoelectric media with arbitrary anisotropy were derived.
Based on these equations and using finite element method (FEM) the full set of
effective moduli for PZT porous ceramics having wide porosity range was
calculated. Different models of representative volume were considered:
piezoelectric cubes with one cubic and one spherical pore inside, cubic volume
evenly divided on partial cubic volumes a part of which randomly declared as
pores etc. For the modeling of the porous piezoceramics with 3-3 connectivity
the representative volume having skeleton structure was considered.
For accounting of inhomogeneous or incomplete ceramics polarization the
preliminary modeling of polarization process was performed. To determine the
areas having different polarization FEM calculations of the electrostatic problem
were executed.
The results of FEM modeling were compared with the theoretical results obtained
on the basis of Marutake’s effective medium approximation, Bruggeman’s
formulas and unit-cell models, as well as with the experimental results for different
porous ceramics in the relative porosity range of 0-70%. Based on these results

138
the modeling of high intensity focusing transducers made of “hard” porous PZT
ceramics was performed.

P1M-5
THE TEMPERATURE INFLUENCE ON THE
PIEZOELECTRIC TRANSDUCER NOISE,
MEASUREMENTS AND MODELING.
F. COUTARD*, E. TISSERAND, and P. SCHWEITZER, LIEN, Université Henri
Poincaré, Vandoeuvre, France.
Corresponding e-mail: [Link]@[Link]
This work, developed at the Electronic Laboratory of Instrumentation of Nancy
(LIEN-France), concerns the characterisation of the electrical impedance and
the generated noise of piezoelectric transducers. Special attention is given to
the medium of propagation and the temperature on the noise level. Theoretical
explanations are proposed to justify the relative thermal stability of the noise
level.
This paper is composed of four parts. The first one considers measurements of
the complex impedance and the spectral noise density of the piezoelectric
transducers and are carried out in water and in air. All measurements are made
using the impedance analyser HP4195A and the spectrum analyser Advantest
R3131A. Because the noise level of the sensor is very low, we insert an amplifier
between the analyzer and the sensor. This amplifier increases the level of signal
above the noise floor and adapts the impedances The transducers are made of
PZT and their resonance frequencies are situated around 2 MHz. The
measurement are realised at a temperature fixed to 298 K. We measure the
correlations between the noise spectrum and the impedance.
In the second part, we study the influence of the temperature T on the impedance
and the noise spectrum for square transducers in [Link] temperature range is
298 to 343 K. In this range, the magnitude of the impedance peak decreases by
0.3 % per K and its relative frequency shifts by 186 ppm per K. The temperature
seems to have a little influence on the noise level : the peak magnitude is nearly
constant, with a sensibility coefficient equal to 0.08 % per K.
The third part is reserved to suggestions of theoretical interpretations of the
observed results. On one hand, the decreasing impedance is explained by the
thermal sensibility of certain characteristics of the piezoelectric material such
as the dielectric constant. On the other hand, we show that whatever is the
origin of the noise delivered by the transducer (intrinsic or noise introduced by
the medium), a Johnson noise explains the quasi-stability of the measurements.
If we consider that the noise is generated by the resistive part R of impedance,
the noise DSP is proportional to RT. This DSP remains constant because the
relative decrease of the resistive part (-13.5 %) compensates the relative increase
of temperature (+15 %).
In the last part, we propose a electroacoustic model (type Redwood) which
includes an estimate of the transducer noise. Currently, this model implemented

139
under PSPICE allows to evaluate the signal-to-noise ratio of ultrasonic émission/
réception system.

P1M-6
NONLINEARITIES AND HYSTERESIS PHENOMENA IN
RECIPROCAL ULTRASOUND SYSTEMS.
M. WILLATZEN*, L. WANG, and Y. FENG, Mads Clausen Institute, University of
Southern Denmark, Grundtvigs Alle 150, DK-6400 Sonderborg, Denmark.
Corresponding e-mail: willatzen@[Link]
Results based on a one-dimensional mathematical model of reciprocal ultrasound
systems [1, 2] including nonlinearities [3] and hysteresis phenomena are
presented. In particular, we consider the constitutive relation for the electric
field E to involve, in general, nonlinear terms to third and fifth order in the electric
displacement D while linearity in the strain is assumed. The transmitter system
consists of a single-layer PZT-5 piezoceramic material layer connected in series
to an electrical impedance and a voltage generator. The receiver circuit consists
of a single-layer PZT-5 piezoceramic material (similar parameters as for the
transmitter) and an electrical impedance (the voltage over the the electrical
impedance is the receiver voltage). Results for the E-D curve are presented
subject to a given voltage-input scenario and hysteresis phenomena are
[Link] also present step-response curves for the transmitter-aperture
pressure and the receiver voltage for three cases of constitutive relations: (1)
linear constitutive E-D relationship, (2) linear and third-order nonlinear terms in
the constitutive E-D relationship, and (3) linear, third-order, and fifth-order
nonlinear terms in the constitutive E-D relationship. The present model allows
for acoustic and dielectric losses in the piezoceramic layers and the possibility
to interconnect a fluid/solid medium between the two PZT transducers.
[1] M. Willatzen, “Ultrasound Transducer Modeling - General Theory and
Applications to Ultrasound Reciprocal System,” IEEE Transactions on Ultrasonic,
Ferroelectrics and Frequency Control, Vol. 48, No.1 January 2001.
[2] M. Willatzen, “Ultrasound transducer modeling - Received voltage signals
and the use of half-wavelength window layers with acoustic coupling layers,”
IEEE Transactions on Ultrasonics, Ferroelectrics, and Frequency Control, 46,
pp. 1164-1174, September 1999.
[3] L. Wang and M. Willatzen, “Mathematical Modelling of Reciprocal Transducer
Systems Accounting for Nonlinear Constitutive Relations,” submitted to
Ultrasonics.

P1M-7
PSEUDOSPECTRAL TIME-DOMAIN METHOD TO
CALCULATE RADIATION PATTERN OF LENS-
FOCUSED TRANSDUCERS.
C. BATIFOL, S. CALLÉ*, P. MARÉCHAL, M. LETHIECQ, and F. LEVASSORT,
François Rabelais University, LUSSI, FRE 2448 CNRS, Tours Cedex, France.
Corresponding e-mail: calle_s@[Link]
140
In ultrasound applications, such as medical imaging or therapy, focusing effect
is used to improve both sensitivity and lateral resolution. This is often obtained
by the addition of an acoustic lens on the transducer as a final element, in
contact with the propagation medium. Previous study on this subject using finite
element method (FEM) and an extended unidimensional KLM model [1] shows
that the initial assumed plane wave source delivers an heterogeneous spectrum
at the lens surface, resulting from longitudinal and transverse waves propagation
and also multiple reflections in this lens. The purpose of this study is to describe
a pseudospectral time-domain (PSTD) algorithm which takes into account this
phenomenon to calculate the radiation pattern of a single-element lens-focused
transducer. This method has the main advantage to follow easily the temporal
evolution of the displacement or pressure propagated fields, in particular through
the lens, and allows to clearly show the different phenomena occurring in this
lens.
The PSTD algorithm, which solves time-dependent partial differential equations,
combines conventional Fourier pseudospectral method and perfectly matched
layers (PML) to avoid reflections and counter the wrap-around effect from the
FFTs. The temporal evolution is performed using a 4th order Adams-Bashforth
time integrator. The other advantages of the pseudospectral numerical algorithm
compared to models based on finite difference or finite element methods are its
efficiency to model large-scale problems. It has a better numerical stability and
does not require a large number of nodes per wavelength.
To predict the radiated pressure field of a single-element circular transducer,
this alternative pseudospectral method has been extended to the 2D
axisymmetric case. New conditions according to this configuration are detailed.
Validation of this numerical model has been performed for propagation in fluid
(water) using a comparison with analytical diffraction models for spherically
focused transducers.
Moreover, to study the influence of the lens on plane wave propagation, the
model has been updated for wave propagation in an elastic solid. In the particular
case of a lens-focused transducer with a 10 MHz center frequency, the pressure
field is deduced at the lens surface using the pseudospectral method. This result
is compared with those obtained using a finite element method (ATILA) in the
same configuration. At the focal point, properties are also in good agreement
with standard models and experimental results [1].
Finally, the presented modeling is used to study the influence of several acoustical
parameters of the lens such as the acoustical impedance or the shear wave
velocity on the radiation pattern, particularly on the electro-acoustic response at
the focal point.
[1] P. Marechal, F. Levassort, L.P. Tran-Huu-Hue, M. Lethiecq, Effect of acoustical
properties of a lens on the pulse-echo response of a single element transducer,
IEEE International Ultrasonics Symposium , pp. 1651-1654, 2004.

141
 ORAL SESSIONS

Session: 1A
LOCAL DRUG DELIVERY UCA/THERAPEUTIC
Chair: P. Burns
University of Toronto

1A-1 11:30 a.m.

MECHANISMS OF ULTRASONICALLY-MEDIATED
DRUG DELIVERY: HIGH-SPEED CAMERA
OBSERVATIONS OF MICROBUBBLES WITH
ATTACHED NANOBEADS.
C.T. CHIN*1, A. VAN WAMEL2, M. EMMER2, N. DE JONG2, C. S. HALL1, and A
L. KLIBANOV3, 1Philips Research, New York, 2Erasmus MC, Rotterdam, The
Netherlands, 3University of Virginia, Charlottesville, VI.
Corresponding e-mail: [Link]@[Link]
Background: Ultrasonically-triggered drug delivery may be accomplished through
the use of microbubble agents modified to carry pharmaceuticals. Understanding
of the dynamics and mechanisms of drug release is needed to ensure robust
and reproducible delivery of pharmaceutical payloads. The objective of this study
was to observe the behavior of lipid-encapsulated microbubbles with attached
fluorescently labeled nanobeads within an ultrasonic field.
Methods: Stabilized perfluorobutane microbubbles were prepared by self-
assembly of a lipid monolayer shell on the gas-liquid interface; aqueous
dispersion of phosphatidylcholine, PEG stearate and biotin-PEG-
phosphatidylethanolamine was subjected to 20 kHz ultrasound in the presence
of perfluorobutane gas. Microbubbles were purified by flotation. Fluorescent
avidin beads were immobilized on the surface of biotinylated bubbles, forming
pendant-like structures. The microbubble-bead complexes were immobilized
on Petri dish surface and monitored with an epifluorescence microscope
equipped with a 40× water immersion objective with a high speed video camera
operated at 500 frames/sec (fps). Microbubbles were exposed to an ultrasonic
field from a Philips HDI-5000 scanner (MI = 1.5). Observable nanobeads were
detached from the bubbles within the time scale (msec) of the experiment. Further
detailed observations were made using the Brandaris-128 camera operated at
13 Mfps. The bead-bubble complexes were exposed to ultrasonic pulses (2.0
MHz, 10-cycles, 0.65 MPa peak negative amplitude) from a single-element
transducer. Dynamics of the bubble boundary and nanobeads were tracked
using automatic and manual image processing tools.
Results: Release and displacement of microbubble-associated fluorescent
nanobeads from their original position was observed during insonation.
Movement of the nanobeads showed detachment of beads from the bubble
shell occurred within 2 to 10 cycles of insonation. Detached beads track the
fluid flow induced by the bubble compression and expansion at velocities up to

142
20 m/sec. The beads motion relative to the bubble surface was due to translational
movement of the bubble, the beads or both. The translational and oscillatory
movement of detached beads diminishes as the distance between the bubble
and the bead increases. At distances more than 10-20 µm, detached beads
come to a rest.
Conclusion: This study reveals short time-scale detail of the release of shell-
associated material from microbubbles achieved through insonation. The tracked
movement of nanobeads in the particular experimental conditions studied infers
the mechanism of release may not necessarily involve rapid ejection of the drug
payload at near-sonic velocities as some previous models have implied.
Therefore, the results did not establish the capability of ultrasound to propel
nanobeads to a distance comparable to a few cell layers. Further experiments
in a wider range of conditions will be performed to evaluate the additional
mechanisms of ultrasound-triggered drug delivery using microbubble complexes.

1A-2 11:45 a.m.

MECHANISMS OF CELL MEMBRANE
PERMEABILIZATION WITH ULTRASOUND AND
CONTRAST MICROBUBBLES.
T. A. TRAN*2,3, S. ROGER2,3, J. Y. LEGUENNEC2,3, F. TRANQUART1,4, and A.
BOUAKAZ1,3, 1Inserm, U619, Inserm, U619, Tours, France, 2Inserm, E0211,
Inserm, E0211, Tours, France, 3Université F. Rabelais, Université F. Rabelais,
Tours, France, 4CIT Ultrasons, CIT Ultrasons, Tours, France.
Corresponding e-mail: bouakaz@[Link]
Aim: New clinical applications of ultrasound contrast agents extend beyond
imaging and diagnostic towards therapeutic applications. A number of
experimental findings have now demonstrated evidence of increased cell
membrane permeability through sonoporation process. In this study, we explore
the mechanisms by which the activation of microbubbles with ultrasound waves
breach cell membranes.
Method: An electrophysiological experimental method is set up to explore the
cell membrane response in sonoporation conditions. This experimental method
consists of measuring the variations in membrane potential which directly
indicates the modulation of ion exchange through the cell membrane and thus
its conductance. To measure the cell activity during sonoporation, patch clamp
technique was used in the “whole cell” configuration where the membrane
potential of a single cell is measured. Mammary cancer cells issued from MDA-
MB-231 cell-lines were used. Sonovue microbubbles provided by Bracco
Research were continuously infused to the cells at a rate of 1ml/min. Ultrasound
was applied using single element transducers of 1 and 2.25MHz, both focused
at 14 mm. Waveforms of different lengths (20 to 40 cycles), amplitudes (50 to
400 kPa), repetition rates (100 to 500 µs) and exposure times (5 to 20 sec) were
transmitted. The microbubbles and cells were simultaneously monitored during
ultrasound exposure using a video camera.
Results: The results revealed that during sonoporation, an obvious
hyperpolarization of the cell membrane potential occurs during the ultrasound

143
excitation, indicating the triggering of specific ion channels while the cell and
the bubble remain both intact. At the highest acoustic exposure (pulses of 40
cycles at 1 MHz repeated every 100 µs during 20 sec); the membrane potential
varied from -30 mV (resting value) to -60 mV and this phenomenon was entirely
reversible. This mechanism showed to be dependent on the number of contrast
microbubbles in the close vicinity of the cell, but revealed that only cells in direct
contact with the bubbles undergo membrane hyperpolarization. Smaller acoustic
amplitudes or higher frequency induced only mild hyperpolarization (less than
20mV) while shutting off the ultrasound brings the potential to its resting value.
However ultrasound alone does not affect the cell membrane potential.
Conclusions: The results demonstrate that microbubble’s oscillations under
ultrasound activation entail modifications of the electrophysiological cell activities,
by triggering the modulation of ionic transports through the plasmic cell
membrane. However, only cells in a direct contact with the microbubbles are
impacted. The involved mechanisms are related to activation of specific channels
sensitive to mechanical stresses (stretch-activated channels) and possibly non-
specific ion channels.

1A-3 12:00 p.m.

TRAPPING OF LOW SENSITIVITY OBJECT
BY SEED BUBBLES.
Y. YAMAKOSHI*, Gunma University, Kiryu, Gunma, Japan.
Corresponding e-mail: yamakosi@[Link]
An ultrasonic wave assisted micro object manipulation is a powerful tool, for
example, in future Drug/Gene delivery system. Trapping of micro vehicles against
blood flow is a targeting technology at desired part. Sonoporation, which makes
micro holes through cell membrane by bubble explosion, improves dose to the
tissue. In these methods, it is assumed that micro bubbles have enough sensitivity
to the ultrasonic waves. It is required that the micro bubbles show large volumetric
oscillation so that they receive enough acoustic radiation force for the bubble
trapping. It is also supposed that the bubbles explode to make jet flow as a
result of nonlinear oscillation. However, these features are not often seen under
an actual situation. For example, payloads of bubble harden the bubble shell
resulting in decrease of volumetric oscillation. Stiff bubble is difficult to explode.
A solution to this problem is to use two kinds of bubbles, which have their own
functions. First bubble is a seed bubble, which has higher sensitivity to the
ultrasonic wave though it does not carry any payloads. The second bubble is a
target bubble, which is an object vehicle though it has lower sensitivity. In this
paper, trapping of low sensitivity object by seed bubble is proposed. First, only
the seed bubbles are introduced into the ultrasonic wave field. The seed bubbles
make bubble aggregation and the aggregated bubbles are trapped. Then, if
target bubbles are introduced, the spatial restricted Bjerknes force field, which
is produced by the seed bubbles, traps the target bubbles to make bi-layered
bubbles mass (Inner layer: seed bubbles, outer layer: target bubbles). Feature
of the method is that the target bubbles are trapped under maximum force
condition while they are trapped under balance of force condition (zero force
condition) in the conventional ultrasonic wave standing wave method. Bubble

144
mass, which is produced under maximum force condition, is stable for additional
forces by, for example, flow turbulence. Second feature is that the produced
condensed mass of seed bubble and target bubble is useful in seed bubble
based sonoporation, because payloads of the target bubbles are placed by
adjoining to the seed bubbles, which are easy to explode under ultrasonic wave
radiation. Experiments are carried out using Levovist (Schering A.G.) as a seed
bubble and Saccharomyces yeast cells (about 5 µm in size) as an example of
target object. It was already proposed that Yeast cell is a vehicle of Drug/Gene
delivery system applied in GI tract. Yeast cell is not a bubble but it has some
sensitivity to the ultrasonic wave, which may come from micro bubbles produced
by fermentation. Ultrasonic wave frequency is from 2.5MHz to 7.5 MHz. The
sound pressure is 100 kPa. Flow velocity is set to 5 mm/sec. Three different
conditions of yeast cell are examined. It was confirmed experimentally that yeast
cells are not trapped when only the yeast cells are introduced into the ultrasonic
wave field. However they are trapped as cloud around the aggregated seed
bubbles, especially when the yeast cells are prepared in the water of temperature
75 degrees in advance.

1A-4 12:15 p.m.

ULTRASOUND-INDUCED UPTAKE OF DIFFERENT
SIZE MARKERS IN MAMMALIAN CELLS.
R. KARSHAFIAN*1,2, S. SAMAC2, M. BANERJEE2, P. D. BEVAN1,2, and P. N.
BURNS1,2, 1Medical Biophysics, University of Toronto, Toronto, ON, Canada,
2
Imaging Research, SWCHSC, Toronto, ON, Canada.
Corresponding e-mail: raffik@[Link]
The delivery of impermeable compounds to cells when exposed to ultrasound
have been demonstrated both in in-vitro and in-vivo studies. The physical
mechanisms by which this occurs have yet to be elucidated. However,
mechanisms due to bubble disruption (formation of microjets and shockwaves)
and stable bubble oscillation (acoustic microstreaming) in ultrasonic fields are
almost certainly implicated. One of the bioeffects is the formation of pores on
the surface of cells. Understanding this process is of vital importance in
establishing parameters for drug delivery in-vivo. We hypothesize that the pore
size formed in cell membrane depends on bubble disruption phenomenon, with
lower and higher acoustic pressures causing smaller and larger pores,
respectively. Here, we use a suspension of cells and calibrated ultrasound beams
to investigate the uptake of different size markers at various exposure parameters
in the absence and presence of microbubbles.
Experiments were carried out on a suspension of cells (KHT-C: murine
fibrosarcoma cell line) exposed to a calibrated ultrasound beam, at various
acoustic pressures (0 to 1MPa) and frequencies (0.5 and 1MHz), in the absence
and presence of microbubbles (7% volume Optison). FITC-dextran was used
as a permeability marker with molecular weights of 10, 40, 70 and 500kDa to
which a cell membrane is normally impermeable. Flow cytometry and microscopy
were used to identify and quantify cell permeability and viability.
The presence of microbubbles is crucial in causing detectable changes in cell
membrane permeability; no permeability changes were observed in the absence

145
of microbubbles even at high pressures (Pneg 1MPa). The sonoporation of the
KHT-C cells is a pressure threshold dependent phenomenon. No effect was
seen at pressures below 150kPa. The uptake of different molecular weight
markers showed similar peak negative threshold values. At lower pressures
(<200kPa), the same percentage of cells (6-8%) showed an uptake of FITC-
dextran, regardless of the marker size. At higher pressures (>400kPa), however,
more cells showed an uptake of the 10kDa (25%) than of the larger markers
(13%). The number of permeabilized cells increased with peak negative pressure
up to 400kPa, above which a plateau was attained for the higher FITC-dextran
sizes.
We hypothesized that bubbles disrupted at low and high pressures would produce
smaller and larger pores, respectively. The results indicate that large pores are
produced when bubbles are disrupted at low pressures. At higher pressures,
more cells showed an uptake of the 10kDa marker only, indicating that small
pores were formed in additional cells. The dominating mode of uptake is
considered to be passive, although, an increase in the active endocytosis could
explain the higher uptake of the 10kDa. Our results imply that the conditions of
the ultrasound exposure must be optimized for a range of drug sizes. Future
work will continue to characterize the effect of different exposure parameters
and bubble combinations on the uptake of bigger size markers.
CIHR

1A-5 12:30 p.m.

(Invited)
TARGETED ULTRASOUND IMAGING
AND DRUG DELIVERY.
K. FERRARA*, University of California, Davis, CA.
Corresponding e-mail: kwferrar@[Link]
Molecularly-targeted ultrasound contrast agents and drug delivery vehicles are
now in pre-clinical evaluation for applications that include cancer and
cardiovascular disease. Here, we review the structure and properties of targeted
contrast agents and results of recent performance characterizations. In addition,
several classes of drug delivery vehicles are under development including gas
bubbles with polymer, lipid, or oil shells, and liquid-filled nanoparticles. High
speed photography is used to characterize the oscillations of contrast agents
and delivery vehicles with and without a targeting ligand. Ultrasound pulsing is
shown to manipulate these particles changing their distribution in vitro and in
vivo and improving the transfer of the drug to nearby cells. Selective targeting to
cells expressing a specific signature will be demonstrated. Ultrasound can also
change the permeability of capillaries within a region, enhancing the delivery of
a drug to the desired local region, and the mechanisms associated with the
changes will be described.
NIH CA 76062, CA 103828, EB 002952

146
Session: 2A
STATIC ELASTICITY
Chair: H. Ermert
Ruhr University

2A-1 11:30 a.m.

A SINGLE-ELEMENT FOCUSED TRANSDUCER
METHOD FOR HARMONIC MOTION IMAGING.
C. MALEKE, M. PERNOT, and E. KONOFAGOU*, Columbia University, New
York, NY.
Corresponding e-mail: ek2191@[Link]
Previously, we demonstrated the feasibility of a technique for simultaneous
monitoring and generation of ultrasound therapy treatment using two separate
focused ultrasound transducer elements [1]. In this study, a new harmonic motion
imaging (HMI) technique is developed that estimates tissue displacement induced
by harmonic radiation force excitation using a single focused ultrasound element.
First, wave propagation simulation models were used to compare the use of
one Amplitude-Modulated (AM) focused beam versus two overlapping focused
beams as previously implemented for HMI [1]. Simulation results indicated that,
unlike the two-beam configuration, the AM beam produced a consistent, stable
focus for the applied harmonic radiation force. The AM beam thus offered the
unique advantage of sustaining the application of the radiation force on the
same tissue region. Experiments were then performed on agar gel phantoms
and ex-vivo tissues. The radiation force was generated by a 4.68 MHz focused
transducer using a low-frequency Amplitude-modulated (AM) RF-signal. A 7.5
MHz single-element, diagnostic transducer was placed through the center of
the focused transducer so that the diagnostic and focused beams were aligned.
A bandpass analog filter was used to remove the spectrum of the focused beam.
Consecutive RF signals were acquired with a PRF of 1 kHz and the displacements
were estimated using a classical cross-correlation algorithm. Absorption
coefficients were measured accordingly. The intensity of the focused beam was
varied from 26.34 W/cm 2 to 421.44 W/cm 2 allowing us to detect tissue
displacements of up to 50 microns with the AM frequency varying between 1 Hz
and 100 Hz. The local elastic properties were reconstructed using a Voigt model
[2], and the regional elastic moduli were found to be between 680 Pa and 1.3
kPa. Finally, taking advantage of the real-time capability of our method, the
change in the elastic properties was monitored during focused ultrasound (FUS)
ablation of ex-vivo tissues. In this experiment, the intensity of the focused
ultrasound beam was 948.23 W/cm2 at the focus, in order to simultaneously
generate the harmonic radiation force, and the tissue temperature variation and
ablation. A decreasing slope of the oscillatory displacements was observed with
time due to the change in the speed of sound temperature. Based on the
harmonic displacements, their temperature-dependence, and the calculated
acoustic radiation force, the change in the regional elastic modulus was monitored
during heating. In conclusion, the feasibility of using an AM radiation force for
harmonic motion imaging for simultaneous monitoring of tissue elasticity variation

147
during ultrasound therapy was demonstrated in phantoms and tissues in vitro.
Further study of this method will include stiffness and temperature mapping, ex
vivo and in vivo.
[1] Konofagou EE, Hynynen K., Ultras Med Biol; 29(10), 1405-1413 (2003).
[2] Krouskop, T., et al., Ultrasonic Imaging 20, 260-274 (1998).
This study was supported by a Special Development Award from the Whitaker
Foundation.

2A-2 11:45 a.m.

RECONSTRUCTION OF THERMAL PROPERTY
DISTRIBUTIONS - THERMAL CONDUCTIVITY,
DIFFUSIVITY, CAPACITY.
C. SUMI* and H. YANAGIMURA, Sophia University, Tokyo, Japan.
Corresponding e-mail: c-sumi@[Link]
For various soft tissues, we are developing the ultrasonic strain measurement-
based mechanical properties (shear modulus, Poisson’s ratio, density)
reconstruction/imaging technique. On human in vivo liver, and breast etc., we
reported the effectiveness as a differential diagnosis technique and a monitoring
technique of low invasive therapies such as chemotherapy, cryotherapy, and
thermal therapy. On these tissues, we also confirmed the inhomogeneity of
thermal-elasticity, for instance, between parenchyma and blood vessels [Trans.
on UFFC. (in press)]. Then, we proposed to reconstruct thermal properties such
as thermal conductivity, thermal diffusivity, and thermal capacity [1] in addition
to these mechanical properties for diagnosis and treatment (planning).
Temperature can be measured by ultrasonic imaging and MRI. Provided that
the reference values of thermal properties are given in the ROI, by solving the
simultaneous first order PDEs (heat transfer equations) having heat fluxes as
inhomogeneous coefficients, we can determine the thermal properties
distributions. To stabilize reconstruction, regularization is simultaneously applied
to each thermal property distribution.
The feasibility was confirmed through simulations and phantom experiments. In
simulations, the 3D techniques were evaluated using a cubic phantom (50 mm
sides, 36 degrees centigrade) containing a central spherical region (dia., 10
mm) which had twice as high conductivity k and specific heat c as those of the
surrounding medium, i.e., 1.0 vs. 0.5 W/(m K), and 8,400 vs. 4,200 J/(K Kg).
The density ρ was uniformly 1,000 Kg/m3. The temperature of the one surface
was changed by 12 degrees centigrade in step. The time series of temperature
distribution were calculated by the successive over relaxation method. Two cases
are considered, i.e., Case 1: two pairs of subsequent temperature distributions
of different time were used; Case 2: two paired subsequent temperature
distributions having opposite heat fluxes were used. In both cases, the thermal
properties were quantitatively reconstructed. Evaluated mean values (spherical
region) of thermal properties are shown in table.
For the phantom experiments, an agar phantom was used (3.87 percent; height,
50 mm). A cylindrical inhomogeneous region (dia., 15 mm) was made at the

148
depth 25 mm using small amount of copper powder. Temperature was measured
using our previously developed phase matching method [Trans. on UFFC., vol.
46, pp. 158-166, 1999] by measuring generated strains due to thermal effect
[Ebbini]. The phantom was heated by hot water from below side. The
inhomogeneous region was detected in respective images of thermal conductivity,
diffusivity, capacity [mean value (6.5 × 5.8 mm) vs. ref. value: conductivity, 0.700
vs. 0.625 W/(m K); diffusivity, 2.20 vs. 1.49 × 10-7 m2/s; capacity, 3.18 vs. 4.20 ×
106 J/(m3 K)].
The demonstrated ability of reconstruction in simulations and phantom
experiments is concluded to confirm the potential value of the developed robust
reconstruction method.
[1] Sumi, Proc Int Conf on the Ultrason Meas and Imag of Tissue Elasticity, p.
73, 2004.
Evaluated mean values of thermal properties (spherical region).

Case k [W/(m K)] rc (´ 106) [J/(m3K)] k/rc (´ 10-7) [m2/s]

Original 1.00 8.40 1.19

1 1.01 8.59 1.18
2 1.05 9.01 1.17

2A-3 12:00 p.m.

DYNAMIC SIGNAL ARRIVAL CORRECTION FOR
VIBRO-ACOUSTOGRAPHY IMAGE FORMATION.
M. URBAN* and J. GREENLEAF, Department of Physiology and Biomedical
Engineering, Mayo Clinic College of Medicine, Rochester, MN.
Corresponding e-mail: [Link]@[Link]
Background: Vibro-acoustography is a recently developed method that uses
the radiation force of ultrasound to locally excite an object. Images are formed
from the resulting acoustic emission measured by a hydrophone that is in close
proximity to the excitation site. In current image formation techniques, we
calculate a pixel’s value by first applying a static temporal window to the recording
of the radiofrequency (RF) signal from the hydrophone, and then calculating
some measure such as a root-mean-square of the windowed data. From our
experience it has been found that the positioning of the hydrophone relative to
the object can affect the overall image contrast because the acoustic emission
has to travel a finite distance to the hydrophone from each excitation point. The
most common way this artifact manifests itself is as a multiplicative illumination
field across the image. Methods: To address this artifact a new algorithm has
been devised that estimates the relative time delay to the hydrophone for each
pixel in the image using an approach based on the Hough transform. We use
the shapes of the wavefronts of the acoustic emission data to estimate correction
angles for two spatial dimensions, denoted as θy and θz. Using these two angles,

149
we create a time delay image that allows us to adjust the location of the temporal
window applied to the RF signal in a dynamic fashion. The algorithm finds the
maximal response of the acoustic emission data and centers the window about
that response and performs the delay correction for each pixel. Results: The
algorithm has been tested using computer phantoms by varying θy and θz in the
ranges -65° ≤ θy ≤ 65° and -65° ≤ θz ≤ 65° in 5° steps, and results show that the
mean squared error for the reconstruction using the dynamic window is in almost
all cases (98.4%) better than the reconstruction using the static window. Image
results will be shown from breast phantom and in vivo breast data. The corrected
images of the breast phantom show improved contrast of lesions because of
the reduction of the multiplicative illumination field. The contrast in the images
is calculated using 1/N S Iβ(m,n) where β = 2 and N is the number of pixels in the
image. For breast phantom data, images formed using the dynamic window
compared with images formed with different static windows used in current
practice show improvement in the image contrast. The percent difference in
contrast ranges from 12% to 200%. In in vivo data, the contrast improvement in
terms of percent difference ranges from 1% to 52%. Conclusions: This method
correctly chooses the segment of the acoustic emission data to process so that
the data is consistent from pixel to pixel despite finite differences in propagation
time to the hydrophone. Results using simulated and real data show that image
contrast and overall image quality is improved with this algorithm.
This study was supported in part by grants EB002640 and EB002167 from the
National Institutes of Health.

2A-4 12:15 p.m.

EVALUATION OF ARRAY SIGNAL PROCESSING
METHODS TO ULTRASOUND-BASED ARTERIAL
PULSE WAVE VELOCITY MEASUREMENTS ON IN
VITRO AND IN VIVO DATA.
A. DENTINGER*, R. HOCTOR, and K. THOMENIUS, GE Global Research,
Niskayuna, NY.
Corresponding e-mail: dentinge@[Link]
The authors have recently developed a model-based signal processing approach
to ultrasound PWV estimation and have shown that this method mitigates the
effect of arterial reflections on simulated data [1]. Initial application to in vivo
data produced PWV estimates that were in the proper physiological range but
with estimation variances of 30-40%. The present paper investigates
modifications to the method to reduce estimation variance for real data, and it
describes physical experiments that were conducted to evaluate the new
methods.
To perform the experiments, a test ultrasound probe was constructed consisting
of six 10 MHz phased arrays, aligned so as to form parallel image planes, and
connected in a single cable assembly with multiplexing electronics. A GE Vivid
7 ultrasound scanner was used to control the probe, enabling interleaved color
M-mode data to be collected from the six arrays. The data was processed off-
line to produce waveforms corresponding to the vessel’s rate of distension. A

150
test tank utilizing a latex tube as a simulated artery was constructed to investigate
the sensitivity of the PWV estimation to reflection geometry.
Two PWV estimation methods were evaluated: the least-squares approach
described in [1] and one based on eigenanalysis of the spatial covariance matri