Neuroscience Letters 456 (2009) 11–14

Contents lists available at ScienceDirect

Neuroscience Letters
journal homepage: [Link]/locate/neulet

A Symbol Digit Modalities Test version suitable for functional MRI studies
Cristina Forn a,∗ , Vicente Belloch b , Juan Carlos Bustamante a , Gabrielle Garbin a ,
María Àntonia Parcet-Ibars a , Ana Sanjuan a , Noelia Ventura a , César Ávila a
a
Dept. Psicología Bàsica, Clínica i Psicobiología, Universitat Jaume I, Castelló, Spain
b
Servicio de Radiología: ERESA, Valencia, Spain

a r t i c l e i n f o a b s t r a c t

Article history: The Symbol Digit Modalities Test is an easy test used to assess cognitive impairment in a wide range of
Received 12 January 2009 neurological diseases, like multiple sclerosis. We adapted the oral version of this cognitive task making it
Received in revised form 24 March 2009 suitable for functional Magnetic Resonance Imaging studies. Symbol Digit Modalities Test performance
Accepted 24 March 2009
was associated with increased brain activity in frontal and parietal areas involved in selective attention and
working memory functions. These may provide the basis for future studies assessing potential abnormal
Keywords:
cortical activations in multiple sclerosis patients and other clinical populations.
Symbol Digit Modalities Test (SDMT)
Published by Elsevier Ireland Ltd.
functional Magnetic Resonance Imaging
(fMRI)
Working memory
Selective attention

The Symbol Digit Modalities Test (SDMT) is a cognitive task devel- N) [13] but it is still suitable for its use in a fMRI experimental
oped by Smith [16]. It consists of a sheet of paper with, at the top, a setup.
sequence of nine symbols and nine corresponding numbers (key). Eighteen right-handed healthy controls (10 female) participated
The task sequence consists of a series of symbols, each with a blank in this study, with a mean age of 32.5 years (S.D. = 6.2) and with a
space underneath. Within a 90-second time limit the subject is mean of 14.5 years of education (S.D. = 2.3). Participants reported
required, consulting the key as necessary, to insert the numbers normal or corrected-to-normal visual acuity and no history of
associated with the symbols. The test can be administered in both medical disabilities. Seven days before scanning, all participants
written and oral modalities. Nowadays, this task is used to assess completed the BRB-N [13] commonly used to assess cognitive
cognitive impairment in a wide range of neurological and neuropsy- impairment in MS patients. An informed consent was obtained
chiatric disorders, especially multiple sclerosis (MS). Indeed, the from them all, and they were monetarily rewarded (50 D ) for their
SDMT has been included as a part of different neuropsychologi- participation in the study.
cal batteries specifically designed to evaluate cognitive deficits in The fMRI task was based on the oral version of the SDMT
this clinical population [2,13]. Further, recent studies have observed included by Rao [13] in the BRB-N. Through scanner-compatible
that SDMT scores are highly correlated to Magnetic Resonance goggles (VisuaStim Resonance Technologies, Inc.) and the Quick
Imaging (MRI) derived measures of disease burden in MS patients Time Player (version 7.1.6; Apple, Inc.) program, the key (an
[12]. upper row containing a sequence of nine symbols and a lower
Recently there has been a notable interest in developing SDMT- row containing nine corresponding numbers) and “probe” stim-
like tasks suitable for its concurrent use in functional MRI (fMRI) uli were presented. “Probe” stimuli were numbers (1–9) in
procedures [6,10,14]. However, these tasks notably differ from the the so-called “control” condition, or meaningless symbols to
original SDMT in several aspects, such as the kind of stimuli pre- be matched to the corresponding numbers in the activation
sented [10] or the required response [6,10,14]. Trying to surpass condition. Both conditions were alternately presented up to a
some of these drawbacks, in the present study we introduced an total of 12 blocks (block duration: 30 s). In each block 15 stim-
adaptation of this task that is closer to the oral version of the SDMT uli were randomly presented during 2 s. This presentation pace
included in the Brief Repeatable Battery-Neuropsychological (BRB- was calculated taking into account the standardized normative
data of the oral version of the SDMT obtained from [3], which
establishes in 1.8 s the average response latency in healthy volun-
∗ Corresponding author at: Dept. Psicología Bàsica, Clínica i Psicobiologia, Campus teers. This task is free for download from [Link]
Riu Sec, Fac. Ciències Humanes i Socials, Universitat Jaume I, E-12071 Castelló, Spain. [Link]. On the other hand, in order to avoid a possible
E-mail address: forn@[Link] (C. Forn). memorization effect, the key containing the symbol–numbers cor-

0304-3940/$ – see front matter. Published by Elsevier Ireland Ltd.

doi:10.1016/[Link].2009.03.081
12 C. Forn et al. / Neuroscience Letters 456 (2009) 11–14

Table 1
Location of brain regions showing significant activation in 18 healthy volunteers during Symbol Digit Modalities Test (SDMT) (p < 0.05, FDR corrected, threshold (k) of 5 voxels
per cluster).

Regions of interest/Brodmann areas (BA) Talairach coordinates

x y z t Voxel

Right occipital lingual gyrus BA 17 16 −87 6 9.22 515

Right occipital lingual gyrus BA 18 30 −70 3 8.96
Right occipital lobe cuneus/BA 17 20 −81 10 9.15
Right occipital lobe fusiform gyrus/BA 19 34 −64 −2 10.09

Left lingual gyrus BA 17 −14 −85 1 9.82 514

Left middle occipital gyrus BA 19 −28 −81 8 10.06 510

Left lingual gyrus BA 17 −20 −82 4 7.72

Left cerebellum posterior lobe, decline −2 −75 −18 9.68 498

Right occipital lobe cuneus/BA 18 20 −96 14 9.05 474

24 −84 21 8.17
Right middle occipital gyrus/BA 18 20 −83 17 8.07

Right occipital lobe cuneus/BA 19 30 −82 24 9.03 471

Right parietal lobe precuneus/BA 7 30 −66 33 8.32 464

Left parietal lobe precuneus/BA 7 −22 −62 56 8.14 456

Left middle temporal gyrus BA 37 −42 −58 0 7.39 444

Left middle frontal gyrus BA 6 −28 0 44 5.36 362

Left occipital lobe cuneus BA 18 −8 −98 19 9.86 352

Left middle occipital gyrus BA 18 −20 −98 25 8.71 247

Left occipital lobe/cuneus BA 18 −12 −98 25 8.65

Left inferior frontal gyrus BA 9 −44 4 31 4.27 215

Left caudate body −2 3 20 3.45 59

Right caudate body 4 2 22 3.45

Right cerebellum, anterior lobe, culmen 30 −58 26 3.41 53

Right caudate tail 20 −26 18 3.26 46

Left cerebellum posterior lobe, semi-lunar −22 −70 −37 3.13 32

Right middle frontal gyrus BA 6 24 0 48 3.23 30

respondence was changed after the 6 first blocks. That is, a different fMRI images were processed with SPM5 (Wellcome Department
association between the same symbols and numbers was employed of Cognitive Neurology, London). After realignment and unwrap-
in the last 6 blocks. This strategy proved to be effective in preventing ping, images were spatially normalized to MNI coordinates by using
memorization effects in a similar study [6]. SPM5 EPI template as reference image, and smoothed with a Gaus-
Subjects were required to produce overt verbal responses that sian kernel (FWHM of 8 mm). Statistical analyses of the group were
were recorded by a researcher located inside the scanner room. performed using the random-effect procedure. The data for each
Thus, during the blocks belonging to the control condition, sub- participant were convolved with the hemodynamic response func-
jects had to overtly repeat the number presented as “probe” stimuli. tion and its time derivative. Motion correction parameters from the
However, during the blocks belonging to the activation condition, realignment were included as regressors at this first level analysis.
subjects had to name the number matched in the key with the sym- Control tasks were compared to the activation task using the one
bol presented as “probe” stimuli. Although according to previous sample t-test (p < 0.05 FDR corrected) with an extend threshold of 5
studies discrete loud responses production does not yield substan- voxels per cluster. MNI coordinates were transformed in Talairach’s
tial changes in the brain activation pattern [1,7], foam cushioning coordinates using a nonlinear transformation [17].
was used to immobilize the head within the coil and subjects were Subjects displayed a normal execution in all neuropsycholog-
specifically trained to minimize head movements while respond- ical tests according to the cut-off values for Spanish population
ing. Therefore, the required response in this adaptation of SDMT [15], then discarding any cognitive impairment. Performance of the
for fMRI studies was the same requested in the oral version of the fMRI-adapted version of the SDMT task during the scanning process
SDMT included by Rao [13] in the BRB-N. was correct, with subjects exhibiting a mean of 99.01% (S.D. = 1.13)
Participants were scanned with a 1.5 T Siemens Avanto of correct responses. Translational movement parameters never
(Erlange, Germany) using a single-shot gradient-echo EPI sequence exceeded 1 voxel and detected rotations were less than 1◦ in any
(TR = 3000 ms; FOV: 250 mm × 250 mm; matrix = 64 × 64 pixels; direction, subject, or session.
TE = 50 ms; number of echoes: 1; slice thickness: 4.5 mm, no The execution of this adaptation of the SDMT was associated
gap; flip angle: 90◦ ). A total of 29 slices were acquired in with an increase in the activity of several portions of the frontal,
the axial plane parallel to the AC–PC line from bottom to parietal and occipital lobes. Most of these areas showed a similar
top, providing coverage of the entire brain. A morphological activation in both hemispheres. More specifically, the highest lev-
volumetric sagittal 3D T1-weighted fast-field echo sequence els of activity were observed in posterior areas, specifically in the
(TR = 11 ms; FOV 256 mm × 234 mm; matrix = 256 × 224; voxel size bilateral occipital cortex and cuneus. Additionally, inferior parietal
1 mm × 1 mm × 1 mm; TE = 4.9 ms; number of echoes: 1; flip areas were also activated. Frontal activations were focused bilat-
angle = 15◦ ) was also acquired. erally on the middle frontal gyrus and on the left hemisphere and
 C. Forn et al. / Neuroscience Letters 456 (2009) 11–14 13

responses which might be cumbersome in some clinical popula-

tions.
Thus, the SDMT version presented here resembles the oral adap-
tation of this task developed by Smith [16] and Rao [13] more than
any other previous adaptation of this task to the fMRI constrains.
These similarities refer to the characteristics of the stimuli, the cog-
nitive demands and the kind of demanded response. Therefore,
both versions of this task might recruit similar cognitive processes.
Indeed, the execution of this fMRI-suitable version of the SDMT is
associated with an increased activity in the occipital lobes (related
to the perception of visual attention/scanning) and, more notably,
to the functional engagement of a fronto-parietal network which
includes several brain areas previously related to selective atten-
tion and working memory processes [5,8,7,11], which have been
proposed two major cognitive domains assessed by the SDMT [9].
Nevertheless, the SDMT version presented here and the oral adap-
tation of this task notably differs in the stimuli presentation pace.
Thus, the original protocol of the SDMT is self-paced whereas our
task (as well as all previous similar studies [10]) imposed an exper-
imenter predefined inter-stimuli interval, which is independent of
subjects’ performance. This procedural difference is largely depen-
dent on inherent fMRI requirements since paced paradigms results
in tighter control over and potentially reduces head motion during
the scanning process [1], but might also be perceived as a potential
reduction of the usefulness of this task as a measure of information
processing speed, the third major cognitive component measured
by the SDMT. However, we think that testing the same subjects in
this task but using different inter-stimuli intervals could surpass
this limitation without compromising the fMRI-scanning process.
In this regard, the inter-stimuli interval used in this initial study (2 s)
was determined after taking into account the standardized norma-
tive data of the oral version of the SDMT obtained from [3] in healthy
Fig. 1. Brain activity pattern related to the visual version of the SDMT task (p < 0.05;
FDR corrected), with an extend threshold of 5 voxels per cluster. As can be seen the volunteers. Accordingly, we observed that the use of this stimuli
performance of this task resulted in the recruitment of a fronto-parietal–occipital presentation pace resulted in an almost perfect performance of the
network, with Brodmann areas 6, 9, 7, and 17–18–19, exhibiting highest activity. A task on this population, but it could be much more challenging for
significant increase in the activity of several cerebellar areas was observed as well. some neurological patients or for low-educated or aged people.
In agreement with this proposal, it should be noted that previ-
ous studies trying to adapt the SDMT to the fMRI requirements
were located in the inferior frontal gyrus. Finally, bilateral activa- in clinical populations used longer inter-stimuli intervals (e.g. 3 s)
tions were also observed in the caudate nucleus and cerebellum [10].
(see Table 1 and Fig. 1). In summary, we presented a new version of the SDMT task that
The main objective of the present study was to introduce a new might be used in an fMRI experimental setup. This version is closer
version of SDMT suitable for fMRI studies but able to provide clin- to the oral adaptation of this test performed by Smith [16] and
ically relevant information. Accordingly, we tried to keep this new Rao [13] than other previous attempts. The brain activation pattern
version of the SDMT as close as possible to the oral adaptation of the associated with the execution of this version of the SDMT confirmed
SDMT developed by Smith [16] and Rao [13], which is widely used that attention- and working-memory related processes are two
in the cognitive evaluation of different clinical populations. This major cognitive components involved on the performance of this
strategy resulted in major differences towards previous attempts task. Finally, we have to indicate some limitations of this study. First,
to adapt the SDMT to fMRI requirements [6,10,14]. Thus, conversely fMRI scanning imply the use of an experimenter-imposed stimuli
to the study of Grabner et al. [10], we kept the original meaning- presentation pace, then limiting the use of this task when studying
less symbols designed for this task as the presented stimuli. This the inter-individuals’ differences on information processing speed,
methodological decision not only kept our task closer to the original which are relevant to understand the cognitive impairment asso-
SDMT but also avoided putative alterations of the brain activation ciated with some neurological diseases such as MS. However, we
pattern due to the use of faces, which visual processing is quite propose futures studies that the use of different inter-stimuli inter-
different of other stimuli. Further, in the SDMT version presented vals might allow the use of this SDMT as a suitable measure of this
here, subjects were requested to perform a cognitive operation (e.g. cognitive trait. Second, future studies should take into account pos-
matching numbers to those meaningless symbols according to a sible visual acuity disturbances (not only in clinical populations,
given key), which is virtually identical to that demanded in the but also in healthy subjects), since a strong relationship between
original formulation of this test. This is a critical difference with visual acuity and SDMT execution has been previously reported
other previous studies trying to adapt this task to fMRI require- [4].”
ments, in which subjects were asked to make a binary decision
about whether or not a series of numbers–symbol pairs matched Acknowledgements
a provided key [6,14]. Finally, and again in close parallelism to the
SDMT protocol used in the clinical practice [2,16,13], in our adapta- This research was supported by grants provided by the Fun-
tion subjects were asked to orally report their responses, whereas dación Dávalos-Fletcher and the Consolider-Ingenio 2010 program
all previous studies [6,10,14] involved the use of arbitrary motor (Spanish Ministry of Science and Education) to C. Ávila.
14 C. Forn et al. / Neuroscience Letters 456 (2009) 11–14

References patients in the Paced Auditory Serial Addition Test (PASAT), J. Clin. Exp. Neu-
ropychol. 13 (2008) 1–8.
[1] S. Basho, E.D. Palmer, M.A. Rubio, B. Wulfeck, R.A. Müller, Effects of generation [9] C. Forn, N. Ventura-Campos, A. Berenguer, V. Belloch, M.A. Parcet, C. Ávila, A
modi in FMRI adaptations of semantic fluency: paced production and overt comparison of brain activation patterns during covert and overt paced auditory
speech, Neuropsychologia 45 (2007) 1697–1706. serial addition test tasks, Hum. Brain Mapp. 29 (2008) 644–650.
[2] R.H. Benedict, J.S. Fischer, C.J. Archibald, P.A. Arnett, W.W. Beatty, J. Bobholz, G.J. [10] R.H. Grabner, F. Popotnig, S. Ropele, C. Neuper, F. Gornai, K. Petrovic, F. Ebner, S.
Chelune, J.D. Fisk, D.W. Langdon, L. Caruso, F. Foley, N.G. LaRocca, L. Vowels, Strasser-Fuchs, F. Fazekas, C. Enzinger, Brain activation patterns elicited by the
A. Weinstein, J. DeLuca, S.M. Rao, F. Munschauer, Minimal neuropsychological ‘Faces Symbol Test’—a pilot fMRI study, Mult. Scler. 14 (2008) 354–363.
assessment of MS patients: a consensus approach, Clin. Neuropsychol. 16 (2002) [11] J.S. Mayer, R.A. Bittner, D. Nikolic, C. Bledowski, R. Goebel, D.E. Linden, Common
381397. neural substrates for visual working memory and attention, Neuroimage 236
[3] J.B. Boringa, R.H. Lazaron, I.E. Reuling, J.J. Adèr, L. Pfenning, J. Lindeboom, L.M. (2007) 441–453.
de Sonneville, N.F. Kalkers, C.H. Polman, The brief repeatable battery of neu- [12] B.A. Parmenter, B. Weinstock-Guttman, N. Garg, F. Munschauer, R.H.B. Benedict,
ropsychological tests: normative values allow application in multiple sclerosis Screening for cognitive impairment in multiple sclerosis using the Symbol Digit
clinical practice, Mult. Scler. 7 (2001) 263–267. Modalities Test, Mult. Scler. 13 (2007) 52–57.
[4] J.M. Bruce, A.S. Bruce, P.A. Arnett, Mild visual acuity disturbance are associated [13] S.M. Rao, A manual for the Brief Repeatable Battery of Neuropsychological Test
with performance on tests of complex visual attention in MS, JINS 13 (2007) in Multiple Sclerosis, National Multiple Sclerosis Society, New York, 1990.
544–548. [14] B. Rypma, J.S. Berger, V. Prabhakaran, B.M. Bly, D.Y. Kimberg, B.B. Biswal, M.
[5] M. Corbetta, G.L. Shulman, Control of goal-directed and stimulus-driven atten- D’Esposito, Neural correlates of cognitive efficiency, Neuroimage 33 (2006)
tion in the brain, Nat. Rev. Neurosci. 3 (2002) 201–215. 969–979.
[6] J. Deluca, H.M. Genova, F.G. Hillary, G. Wylie, Neural correlates of cognitive [15] J. Sepulcre, S. Vanotti, R. Hernández, G. Sandoval, F. Cáceres, O. Garcea, P. Vil-
fatigue in multiple sclerosis using functional MRI, J. Neurol. Sci. 270 (1–2) (2008) loslada, Cognitive impairment in patients with multiple sclerosis using the Brief
28–39. Repeatable Battery-Neuropsychology test, Mult. Scler. 12 (2006) 1–9.
[7] C. Forn, A. Barrós-Loscertales, J. Escudero, V. Belloch, S. Campos, M.A. Parcet- [16] A. Smith, Symbol Digit Modalities Test, Western Psychological Services, Los
Ibars, A. Ávila, Cortical reorganization during PASAT task in MS patients with Angeles, 1982.
preserved working memory functions, Neuroimage 31 (2006) 686–689. [17] J. Talairach, P. Tournoux, Co-planar Stereotaxic Atlas of the Human Brain: 3-
[8] C. Forn, A. Belenguer, M.A. Parcet-Ibars, C. Ávila, Information processing speed Dimensional Proportional System an Approach to Medical Cerebral Imaging,
is the primary deficit underlying the poor performance of multiple sclerosis Thieme, Stuttgart, 1988.