Currently, microalgae (i.e., diatoms) constitute a generally accepted bioindicator of water quali... more Currently, microalgae (i.e., diatoms) constitute a generally accepted bioindicator of water quality and therefore provide an index of the status of biological ecosystems. Diatom detection for specimen counting and sample classification are two difficult time-consuming tasks for the few existing expert diatomists. To mitigate this challenge, in this work, we propose a fully operative low-cost automated microscope, integrating algorithms for: (1) stage and focus control, (2) image acquisition (slide scanning, stitching, contrast enhancement), and (3) diatom detection and a prospective specimen classification (among 80 taxa). Deep learning algorithms have been applied to overcome the difficult selection of image descriptors imposed by classical machine learning strategies. With respect to the mentioned strategies, the best results were obtained by deep neural networks with a maximum precision of 86{\%} (with the YOLO network) for detection and 99.51{\%} for classification, among 80 different species (with the AlexNet network). All the developed operational modules are integrated and controlled by the user from the developed graphical user interface running in the main controller. With the developed operative platform, it is noteworthy that this work provides a quite useful toolbox for phycologists in their daily challenging tasks to identify and classify diatoms.
High-resolution images of phytoplankton cells such as diatoms or desmids, which are useful for mo... more High-resolution images of phytoplankton cells such as diatoms or desmids, which are useful for monitoring water quality, can now be provided by digital microscopes, facilitating the automated analysis and identification of specimens. Conventional approaches are based on optical microscopy; however, manual image analysis is impractical due to the huge diversity of this group of microalgae and its great morphological plasticity. As such, there is a need for automated recognition techniques for diagnostic tools (e.g. environmental monitoring networks, early warning systems) to improve the management of water resources and decision-making processes. Describing the entire workflow of a bioindicator system, from capture, analysis and identification to the determination of quality indices, this book provides insights into the current state-of-the-art in automatic identification systems in microscopy.
Automatic diatom identification approaches have revealed remarkable abilities to tackle the chall... more Automatic diatom identification approaches have revealed remarkable abilities to tackle the challenges of water quality assessment and other environmental issues. Scientists often analyze the taxonomic characters of the target taxa for automatic identification. In this process the digital photographs, sketches or drawings are recorded to analyze the shape and size of the frustule, the arrangement of striae, the raphe endings, and the striae density. In this paper, we describe two new methods for producing drawings of different diatom species at any stage of their life cycle development that can also be useful for future reference and comparisons. We attempt to produce drawings of diatom species using Edge-preserving Multi-scale Decomposition (EMD). The edge preserving smoothing property of Weighted Least Squares (WLS) optimization framework is used to extract high-frequency details. The details extracted from two-scale decomposition are transformed to drawings which help in identifying possible striae patterns from diatom images. To analyze the salient local features preserved in the drawings, the Scale Invariant Feature Transform (SIFT) model is adopted for feature extraction. The generated drawings help to identify certain unique taxonomic and morphological features that are necessary for the identification of the diatoms. The new methods have been compared with two alternative pencil drawing techniques showing better performance for details preservation.
This chapter introduces the antecedents, motivation, and necessity of the use of automatic identi... more This chapter introduces the antecedents, motivation, and necessity of the use of automatic identification methods in diatom taxonomy. Expert biologists have a repetitive and laborious identification mission. The principal taxonomic features used to describe and classify diatoms relate to the morphology and texture of frustule. Classical taxonomic diagnosis is carried out by means of identification keys or by visual comparison with respect to standard preparations or reference iconographies. Automatic diatom identification remains an open challenge because, for instance, many diatoms that have been known by the same species for decades have subsequently been split into different species, while on the other hand the emergence of new species is continuous. The very promising results of the new deep learning techniques together with the development of new optical devices in microscopy allow to predict a significant advance in the field.
Introduction/ Background In recent years different technological solutions have emerged for the s... more Introduction/ Background In recent years different technological solutions have emerged for the scanning or digitization of histological and cytological slides in pathology, from several manufacturers. High resolution scanning is usually based in tile (small fragment) or stripes (longitudinal areas) that are combined or stitched together to create a high magnification (usually equivalent to 20x to 40x magnification) global image. Thus, a large digital slide can be displayed using specific viewers to simulate the functions of a conventional microscope. A pyramid of images is a common solution. But each scanner manufacturer optimizes the process of collecting, managing and storing images in its own format, making difficult the interconnection between systems and the ability to share images between different formats, and, generally a heavy process of image conversion and a loss of information is needed. Aims The objective is to present the process performed on proprietary image formats...
Currently, microalgae (i.e., diatoms) constitute a generally accepted bioindicator of water quali... more Currently, microalgae (i.e., diatoms) constitute a generally accepted bioindicator of water quality and therefore provide an index of the status of biological ecosystems. Diatom detection for specimen counting and sample classification are two difficult time-consuming tasks for the few existing expert diatomists. To mitigate this challenge, in this work, we propose a fully operative low-cost automated microscope, integrating algorithms for: (1) stage and focus control, (2) image acquisition (slide scanning, stitching, contrast enhancement), and (3) diatom detection and a prospective specimen classification (among 80 taxa). Deep learning algorithms have been applied to overcome the difficult selection of image descriptors imposed by classical machine learning strategies. With respect to the mentioned strategies, the best results were obtained by deep neural networks with a maximum precision of 86% (with the YOLO network) for detection and 99.51% for classification, among 80 different...
Automatic Non-rigid Histological Image Registration (ANHIR) challenge was organized to compare th... more Automatic Non-rigid Histological Image Registration (ANHIR) challenge was organized to compare the performance of image registration algorithms on several kinds of microscopy histology images in a fair and independent manner. We have assembled 8 datasets, containing 355 images with 18 different stains, resulting in 481 image pairs to be registered.
Diatoms are unicellular algae present almost wherever there is water. Diatom identification has m... more Diatoms are unicellular algae present almost wherever there is water. Diatom identification has many applications in different fields of study, such as ecology, forensic science, etc. In environmental studies, algae can be used as a natural water quality indicator. The diatom life cycle consists of the set of stages that pass through the successive generations of each species from the initial to the senescent cells. Life cycle modeling is a complex process since in general the distribution of the parameter vectors that represent the variations that occur in this process is non-linear and of high dimensionality. In this paper, we propose to characterize the diatom life cycle by the main features that change during the algae life cycle, mainly the contour shape and the texture. Elliptical Fourier Descriptors (EFD) are used to describe the diatom contour while phase congruency and Gabor filters describe the inner ornamentation of the algae. The proposed method has been tested with a sm...
The introduction of digital pathology to nephrology provides a platform for the development of ne... more The introduction of digital pathology to nephrology provides a platform for the development of new methodologies and protocols for visual, morphometric and computer-aided assessment of renal biopsies. Application of digital imaging to pathology made substantial progress over the past decade; it is now in use for education, clinical trials and translational research. Digital pathology evolved as a valuable tool to generate comprehensive structural information in digital form, a key prerequisite for achieving precision pathology for computational biology. The application of this new technology on an international scale is driving novel methods for collaborations, providing unique opportunities but also challenges. Standardization of methods needs to be rigorously evaluated and applied at each step, from specimen processing to scanning, uploading into digital repositories, morphologic, morphometric and computer-aided assessment, data collection and analysis. In this review, we discuss the status and opportunities created by the application of digital imaging to precision nephropathology, and present a vision for the near future.
Pollen identification is required in different scenarios such as prevention of allergic reactions... more Pollen identification is required in different scenarios such as prevention of allergic reactions, cilmate analysis or apiculture. However, it is a timeconsuming task since experts are required to recognize each pollen grain through the microscope. In this study, we performed an exhaustive ass
An essential and indispensable component of automated microscopy framework is the automatic focus... more An essential and indispensable component of automated microscopy framework is the automatic focusing system, which determines the in-focus position of a given field of view by searching the maximum value of a focusing function over a range of z-axis positions. The focus function and its computation time are crucial to the accuracy and efficiency of the system. Sixteen focusing algorithms were analyzed for histological and histopathological images. In terms of accuracy, results have shown an overall high performance by most of the methods. However, we included in the evaluation study other criteria such as computational cost and focusing curve shape which are crucial for real-time applications and were used to highlight the best practices.
Immunohistochemistry (IHC) plays an essential role in Pathology. In order to improve reproducibil... more Immunohistochemistry (IHC) plays an essential role in Pathology. In order to improve reproducibility and standardization of the results interpretation, IHC quantification methods have been developed. IHC interpretation based in whole slide imaging or virtual microscopy is of special interest. The objective of this work is to review the different computerbased programs for automatic immunohistochemistry and Fluorescence In Situ Hybridization (FISH) evaluation. Scanning solutions and image analysis software in immunohistochemistry were studied, focusing especially on systems based in virtual slides. Integrated scanning and image analysis systems are available (Bacus TMAScore, Dako ACIS III, Genetix Ariol, Aperio Image Analysis, 3DHistech Mirax HistoQuant, Bioimagene Pathiam). Other image analysis software systems (Definiens TissueMap, SlidePath Tissue Image Analysis) can be applied to several virtual slide formats. Fluorescence is the preferred approach in HistoRx AQUA, since it allows for a better compartmentalization of signals. Multispectral imaging using CRi Nuance allows multiple antibodies immunohistochemistry, and different stain unmixing. Most current popular automated image analysis solutions are aimed to brightfield immunohistochemistry, but fluorescence and FISH solutions may become more important in the near future. Automated quantitative tissue microarrays (TMA) analysis is essential to provide high-throughput analysis. Medical informatics standards in images (DICOM) and workflow (IHE) under development will foster the use of image analysis in Pathology Departments.
The goal of this challenge 1 was to evaluate new and existing algorithms for automated detection ... more The goal of this challenge 1 was to evaluate new and existing algorithms for automated detection of metastases in hematoxylin and eosin (H&E) stained whole-slide images of lymph node sections. To this end, both slide-based and lesion-based evaluation was made. Several classification methods were tested including classical and deep learning techniques. The most efficient one for the dataset tested was Bagging Tree classifiers using texture features. In the slide-based classification an AUC equal to 0.9952 was obtained, with 98.13% of TP and 1.28% of FP. The TP result decreases in the lesion-based evaluation. Two methodologies were proposed for this second evaluation. Method 1 was based on the convex area of the regions and method 2 based on morphophonemic, geometric and statistical features. The sensitivity for lesion detection was 39.83% and 36.66% respectively, though the false positive average is kept low about 12.28 in method 1 and 10.71 in method 2.
An essential and indispensable component of automated microscopy framework is the automatic focus... more An essential and indispensable component of automated microscopy framework is the automatic focusing system, which determines the in-focus position of a given field of view by searching the maximum value of a focusing function over a range of z-axis positions. The focus function and its computation time are crucial to the accuracy and efficiency of the system. Sixteen focusing algorithms were analyzed for histological and histopathological images. In terms of accuracy, results have shown an overall high performance by most of the methods. However, we included in the evaluation study other criteria such as computational cost and focusing curve shape which are crucial for real-time applications and were used to highlight the best practices.
Proceedings of the International Conference on Computer Vision Theory and Applications
The complexity in face recognition emerges from the variability of the appearance of human faces.... more The complexity in face recognition emerges from the variability of the appearance of human faces. While the identity is preserved, the appearance of a face may change due to factors such as illumination, facial pose or facial expression. Reliable biometric identification relies on an appropriate response to these factors. In this paper we address the estimation of the facial pose as a first step to deal with pose changes. We present a method for pose estimation from two-dimensional images captured under active infrared illumination using a statistical model of facial appearance. An active appearance model is fitted to the target image to find facial features. We formulate the fitting algorithm using a smooth warp function, namely thin plate splines. The presented algorithm requires only a coarse and generic three-dimensional model of the face to estimate the pose from the detected features locations. The desired field of application requires the algorithm to work with many different faces, including faces of subjects not seen during the training stage. A special focus is therefore on the evaluation of the generalization performance of the algorithm which is one weakness of the classic active appearance model algorithm.
In recent scientific literature, some studies have been published where recognition rates obtaine... more In recent scientific literature, some studies have been published where recognition rates obtained with Deep Learning (DL) surpass those obtained by humans on the same task. In contrast to this, other studies have shown that DL networks have a somewhat strange behavior which is very different from human responses when confronted with the same task. The case of the so-called “adversarial examples” is perhaps the best example in this regard. Despite the biological plausibility of neural networks, the fact that they can produce such implausible misclassifications still points to a fundamental difference between human and machine learning. This paper delves into the possible causes of this intriguing phenomenon. We first contend that, if adversarial examples are pointing to an implausibility it is because our perception of them relies on our capability to recognise the classes of the images. For this reason we focus on what we call cognitively adversarial examples, which are those obtained from samples that the classifier can in fact recognise correctly. Additionally, in this paper we argue that the phenomenon of adversarial examples is rooted in the inescapable trade-off that exists in machine learning (including DL) between fitting and generalization. This hypothesis is supported by experiments carried out in which the robustness to adversarial examples is measured with respect to the degree of fitting to the training samples.
International Journal of Machine Learning and Cybernetics
The phenomenon of Adversarial Examples has become one of the most intriguing topics associated to... more The phenomenon of Adversarial Examples has become one of the most intriguing topics associated to deep learning. The so-called adversarial attacks have the ability to fool deep neural networks with inappreciable perturbations. While the effect is striking, it has been suggested that such carefully selected injected noise does not necessarily appear in real-world scenarios. In contrast to this, some authors have looked for ways to generate adversarial noise in physical scenarios (traffic signs, shirts, etc.), thus showing that attackers can indeed fool the networks. In this paper we go beyond that and show that adversarial examples also appear in the real-world without any attacker or maliciously selected noise involved. We show this by using images from tasks related to microscopy and also general object recognition with the well-known ImageNet dataset. A comparison between these natural and the artificially generated adversarial examples is performed using distance metrics and imag...
Objectives. The present study aimed to examine the sedentary behavior (SB) and physical activity ... more Objectives. The present study aimed to examine the sedentary behavior (SB) and physical activity (PA) levels in children using six selected accelerometry protocols based on diverse cut-off points.Methods. Clinical examination, anthropometric measurements, and PA evaluation by accelerometry were assessed in 543 selected children (10±2.4 years old) from the Spanish GENOBOX study. The ActiLife data scoring program was used to determine daily min spent in SB, and light, moderate, vigorous and moderate-vigorous PA using six validated accelerometry protocols differing in their cutt-off points.Results. Very different estimations for SB and PA intensity levels were found in children, independently of the non-wear-time algorithm selected, and considering puberty stages, age and body mass index. The time spent in daily SB varied from 471 to 663.7 min, PA ranged from 141 to 301.6 min, and the moderate-vigorous PA was reported between 20.7 and 180.2 min.Conclusion. The choice of a particular ac...
Adversarial examples are one of the most intriguing topics in modern deep learning. Imperceptible... more Adversarial examples are one of the most intriguing topics in modern deep learning. Imperceptible perturbations to the input can fool robust models. In relation to this problem, attack and defense methods are being developed almost on a daily basis. In parallel, efforts are being made to simply pointing out when an input image is an adversarial example. This can help prevent potential issues, as the failure cases are easily recognizable by humans. The proposal in this work is to study how chaos theory methods can help distinguish adversarial examples from regular images. Our work is based on the assumption that deep networks behave as chaotic systems, and adversarial examples are the main manifestation of it (in the sense that a slight input variation produces a totally different output). In our experiments, we show that the Lyapunov exponents (an established measure of chaoticity), which have been recently proposed for classification of adversarial examples, are not robust to image...
Currently, microalgae (i.e., diatoms) constitute a generally accepted bioindicator of water quali... more Currently, microalgae (i.e., diatoms) constitute a generally accepted bioindicator of water quality and therefore provide an index of the status of biological ecosystems. Diatom detection for specimen counting and sample classification are two difficult time-consuming tasks for the few existing expert diatomists. To mitigate this challenge, in this work, we propose a fully operative low-cost automated microscope, integrating algorithms for: (1) stage and focus control, (2) image acquisition (slide scanning, stitching, contrast enhancement), and (3) diatom detection and a prospective specimen classification (among 80 taxa). Deep learning algorithms have been applied to overcome the difficult selection of image descriptors imposed by classical machine learning strategies. With respect to the mentioned strategies, the best results were obtained by deep neural networks with a maximum precision of 86{\%} (with the YOLO network) for detection and 99.51{\%} for classification, among 80 different species (with the AlexNet network). All the developed operational modules are integrated and controlled by the user from the developed graphical user interface running in the main controller. With the developed operative platform, it is noteworthy that this work provides a quite useful toolbox for phycologists in their daily challenging tasks to identify and classify diatoms.
High-resolution images of phytoplankton cells such as diatoms or desmids, which are useful for mo... more High-resolution images of phytoplankton cells such as diatoms or desmids, which are useful for monitoring water quality, can now be provided by digital microscopes, facilitating the automated analysis and identification of specimens. Conventional approaches are based on optical microscopy; however, manual image analysis is impractical due to the huge diversity of this group of microalgae and its great morphological plasticity. As such, there is a need for automated recognition techniques for diagnostic tools (e.g. environmental monitoring networks, early warning systems) to improve the management of water resources and decision-making processes. Describing the entire workflow of a bioindicator system, from capture, analysis and identification to the determination of quality indices, this book provides insights into the current state-of-the-art in automatic identification systems in microscopy.
Automatic diatom identification approaches have revealed remarkable abilities to tackle the chall... more Automatic diatom identification approaches have revealed remarkable abilities to tackle the challenges of water quality assessment and other environmental issues. Scientists often analyze the taxonomic characters of the target taxa for automatic identification. In this process the digital photographs, sketches or drawings are recorded to analyze the shape and size of the frustule, the arrangement of striae, the raphe endings, and the striae density. In this paper, we describe two new methods for producing drawings of different diatom species at any stage of their life cycle development that can also be useful for future reference and comparisons. We attempt to produce drawings of diatom species using Edge-preserving Multi-scale Decomposition (EMD). The edge preserving smoothing property of Weighted Least Squares (WLS) optimization framework is used to extract high-frequency details. The details extracted from two-scale decomposition are transformed to drawings which help in identifying possible striae patterns from diatom images. To analyze the salient local features preserved in the drawings, the Scale Invariant Feature Transform (SIFT) model is adopted for feature extraction. The generated drawings help to identify certain unique taxonomic and morphological features that are necessary for the identification of the diatoms. The new methods have been compared with two alternative pencil drawing techniques showing better performance for details preservation.
This chapter introduces the antecedents, motivation, and necessity of the use of automatic identi... more This chapter introduces the antecedents, motivation, and necessity of the use of automatic identification methods in diatom taxonomy. Expert biologists have a repetitive and laborious identification mission. The principal taxonomic features used to describe and classify diatoms relate to the morphology and texture of frustule. Classical taxonomic diagnosis is carried out by means of identification keys or by visual comparison with respect to standard preparations or reference iconographies. Automatic diatom identification remains an open challenge because, for instance, many diatoms that have been known by the same species for decades have subsequently been split into different species, while on the other hand the emergence of new species is continuous. The very promising results of the new deep learning techniques together with the development of new optical devices in microscopy allow to predict a significant advance in the field.
Introduction/ Background In recent years different technological solutions have emerged for the s... more Introduction/ Background In recent years different technological solutions have emerged for the scanning or digitization of histological and cytological slides in pathology, from several manufacturers. High resolution scanning is usually based in tile (small fragment) or stripes (longitudinal areas) that are combined or stitched together to create a high magnification (usually equivalent to 20x to 40x magnification) global image. Thus, a large digital slide can be displayed using specific viewers to simulate the functions of a conventional microscope. A pyramid of images is a common solution. But each scanner manufacturer optimizes the process of collecting, managing and storing images in its own format, making difficult the interconnection between systems and the ability to share images between different formats, and, generally a heavy process of image conversion and a loss of information is needed. Aims The objective is to present the process performed on proprietary image formats...
Currently, microalgae (i.e., diatoms) constitute a generally accepted bioindicator of water quali... more Currently, microalgae (i.e., diatoms) constitute a generally accepted bioindicator of water quality and therefore provide an index of the status of biological ecosystems. Diatom detection for specimen counting and sample classification are two difficult time-consuming tasks for the few existing expert diatomists. To mitigate this challenge, in this work, we propose a fully operative low-cost automated microscope, integrating algorithms for: (1) stage and focus control, (2) image acquisition (slide scanning, stitching, contrast enhancement), and (3) diatom detection and a prospective specimen classification (among 80 taxa). Deep learning algorithms have been applied to overcome the difficult selection of image descriptors imposed by classical machine learning strategies. With respect to the mentioned strategies, the best results were obtained by deep neural networks with a maximum precision of 86% (with the YOLO network) for detection and 99.51% for classification, among 80 different...
Automatic Non-rigid Histological Image Registration (ANHIR) challenge was organized to compare th... more Automatic Non-rigid Histological Image Registration (ANHIR) challenge was organized to compare the performance of image registration algorithms on several kinds of microscopy histology images in a fair and independent manner. We have assembled 8 datasets, containing 355 images with 18 different stains, resulting in 481 image pairs to be registered.
Diatoms are unicellular algae present almost wherever there is water. Diatom identification has m... more Diatoms are unicellular algae present almost wherever there is water. Diatom identification has many applications in different fields of study, such as ecology, forensic science, etc. In environmental studies, algae can be used as a natural water quality indicator. The diatom life cycle consists of the set of stages that pass through the successive generations of each species from the initial to the senescent cells. Life cycle modeling is a complex process since in general the distribution of the parameter vectors that represent the variations that occur in this process is non-linear and of high dimensionality. In this paper, we propose to characterize the diatom life cycle by the main features that change during the algae life cycle, mainly the contour shape and the texture. Elliptical Fourier Descriptors (EFD) are used to describe the diatom contour while phase congruency and Gabor filters describe the inner ornamentation of the algae. The proposed method has been tested with a sm...
The introduction of digital pathology to nephrology provides a platform for the development of ne... more The introduction of digital pathology to nephrology provides a platform for the development of new methodologies and protocols for visual, morphometric and computer-aided assessment of renal biopsies. Application of digital imaging to pathology made substantial progress over the past decade; it is now in use for education, clinical trials and translational research. Digital pathology evolved as a valuable tool to generate comprehensive structural information in digital form, a key prerequisite for achieving precision pathology for computational biology. The application of this new technology on an international scale is driving novel methods for collaborations, providing unique opportunities but also challenges. Standardization of methods needs to be rigorously evaluated and applied at each step, from specimen processing to scanning, uploading into digital repositories, morphologic, morphometric and computer-aided assessment, data collection and analysis. In this review, we discuss the status and opportunities created by the application of digital imaging to precision nephropathology, and present a vision for the near future.
Pollen identification is required in different scenarios such as prevention of allergic reactions... more Pollen identification is required in different scenarios such as prevention of allergic reactions, cilmate analysis or apiculture. However, it is a timeconsuming task since experts are required to recognize each pollen grain through the microscope. In this study, we performed an exhaustive ass
An essential and indispensable component of automated microscopy framework is the automatic focus... more An essential and indispensable component of automated microscopy framework is the automatic focusing system, which determines the in-focus position of a given field of view by searching the maximum value of a focusing function over a range of z-axis positions. The focus function and its computation time are crucial to the accuracy and efficiency of the system. Sixteen focusing algorithms were analyzed for histological and histopathological images. In terms of accuracy, results have shown an overall high performance by most of the methods. However, we included in the evaluation study other criteria such as computational cost and focusing curve shape which are crucial for real-time applications and were used to highlight the best practices.
Immunohistochemistry (IHC) plays an essential role in Pathology. In order to improve reproducibil... more Immunohistochemistry (IHC) plays an essential role in Pathology. In order to improve reproducibility and standardization of the results interpretation, IHC quantification methods have been developed. IHC interpretation based in whole slide imaging or virtual microscopy is of special interest. The objective of this work is to review the different computerbased programs for automatic immunohistochemistry and Fluorescence In Situ Hybridization (FISH) evaluation. Scanning solutions and image analysis software in immunohistochemistry were studied, focusing especially on systems based in virtual slides. Integrated scanning and image analysis systems are available (Bacus TMAScore, Dako ACIS III, Genetix Ariol, Aperio Image Analysis, 3DHistech Mirax HistoQuant, Bioimagene Pathiam). Other image analysis software systems (Definiens TissueMap, SlidePath Tissue Image Analysis) can be applied to several virtual slide formats. Fluorescence is the preferred approach in HistoRx AQUA, since it allows for a better compartmentalization of signals. Multispectral imaging using CRi Nuance allows multiple antibodies immunohistochemistry, and different stain unmixing. Most current popular automated image analysis solutions are aimed to brightfield immunohistochemistry, but fluorescence and FISH solutions may become more important in the near future. Automated quantitative tissue microarrays (TMA) analysis is essential to provide high-throughput analysis. Medical informatics standards in images (DICOM) and workflow (IHE) under development will foster the use of image analysis in Pathology Departments.
The goal of this challenge 1 was to evaluate new and existing algorithms for automated detection ... more The goal of this challenge 1 was to evaluate new and existing algorithms for automated detection of metastases in hematoxylin and eosin (H&E) stained whole-slide images of lymph node sections. To this end, both slide-based and lesion-based evaluation was made. Several classification methods were tested including classical and deep learning techniques. The most efficient one for the dataset tested was Bagging Tree classifiers using texture features. In the slide-based classification an AUC equal to 0.9952 was obtained, with 98.13% of TP and 1.28% of FP. The TP result decreases in the lesion-based evaluation. Two methodologies were proposed for this second evaluation. Method 1 was based on the convex area of the regions and method 2 based on morphophonemic, geometric and statistical features. The sensitivity for lesion detection was 39.83% and 36.66% respectively, though the false positive average is kept low about 12.28 in method 1 and 10.71 in method 2.
An essential and indispensable component of automated microscopy framework is the automatic focus... more An essential and indispensable component of automated microscopy framework is the automatic focusing system, which determines the in-focus position of a given field of view by searching the maximum value of a focusing function over a range of z-axis positions. The focus function and its computation time are crucial to the accuracy and efficiency of the system. Sixteen focusing algorithms were analyzed for histological and histopathological images. In terms of accuracy, results have shown an overall high performance by most of the methods. However, we included in the evaluation study other criteria such as computational cost and focusing curve shape which are crucial for real-time applications and were used to highlight the best practices.
Proceedings of the International Conference on Computer Vision Theory and Applications
The complexity in face recognition emerges from the variability of the appearance of human faces.... more The complexity in face recognition emerges from the variability of the appearance of human faces. While the identity is preserved, the appearance of a face may change due to factors such as illumination, facial pose or facial expression. Reliable biometric identification relies on an appropriate response to these factors. In this paper we address the estimation of the facial pose as a first step to deal with pose changes. We present a method for pose estimation from two-dimensional images captured under active infrared illumination using a statistical model of facial appearance. An active appearance model is fitted to the target image to find facial features. We formulate the fitting algorithm using a smooth warp function, namely thin plate splines. The presented algorithm requires only a coarse and generic three-dimensional model of the face to estimate the pose from the detected features locations. The desired field of application requires the algorithm to work with many different faces, including faces of subjects not seen during the training stage. A special focus is therefore on the evaluation of the generalization performance of the algorithm which is one weakness of the classic active appearance model algorithm.
In recent scientific literature, some studies have been published where recognition rates obtaine... more In recent scientific literature, some studies have been published where recognition rates obtained with Deep Learning (DL) surpass those obtained by humans on the same task. In contrast to this, other studies have shown that DL networks have a somewhat strange behavior which is very different from human responses when confronted with the same task. The case of the so-called “adversarial examples” is perhaps the best example in this regard. Despite the biological plausibility of neural networks, the fact that they can produce such implausible misclassifications still points to a fundamental difference between human and machine learning. This paper delves into the possible causes of this intriguing phenomenon. We first contend that, if adversarial examples are pointing to an implausibility it is because our perception of them relies on our capability to recognise the classes of the images. For this reason we focus on what we call cognitively adversarial examples, which are those obtained from samples that the classifier can in fact recognise correctly. Additionally, in this paper we argue that the phenomenon of adversarial examples is rooted in the inescapable trade-off that exists in machine learning (including DL) between fitting and generalization. This hypothesis is supported by experiments carried out in which the robustness to adversarial examples is measured with respect to the degree of fitting to the training samples.
International Journal of Machine Learning and Cybernetics
The phenomenon of Adversarial Examples has become one of the most intriguing topics associated to... more The phenomenon of Adversarial Examples has become one of the most intriguing topics associated to deep learning. The so-called adversarial attacks have the ability to fool deep neural networks with inappreciable perturbations. While the effect is striking, it has been suggested that such carefully selected injected noise does not necessarily appear in real-world scenarios. In contrast to this, some authors have looked for ways to generate adversarial noise in physical scenarios (traffic signs, shirts, etc.), thus showing that attackers can indeed fool the networks. In this paper we go beyond that and show that adversarial examples also appear in the real-world without any attacker or maliciously selected noise involved. We show this by using images from tasks related to microscopy and also general object recognition with the well-known ImageNet dataset. A comparison between these natural and the artificially generated adversarial examples is performed using distance metrics and imag...
Objectives. The present study aimed to examine the sedentary behavior (SB) and physical activity ... more Objectives. The present study aimed to examine the sedentary behavior (SB) and physical activity (PA) levels in children using six selected accelerometry protocols based on diverse cut-off points.Methods. Clinical examination, anthropometric measurements, and PA evaluation by accelerometry were assessed in 543 selected children (10±2.4 years old) from the Spanish GENOBOX study. The ActiLife data scoring program was used to determine daily min spent in SB, and light, moderate, vigorous and moderate-vigorous PA using six validated accelerometry protocols differing in their cutt-off points.Results. Very different estimations for SB and PA intensity levels were found in children, independently of the non-wear-time algorithm selected, and considering puberty stages, age and body mass index. The time spent in daily SB varied from 471 to 663.7 min, PA ranged from 141 to 301.6 min, and the moderate-vigorous PA was reported between 20.7 and 180.2 min.Conclusion. The choice of a particular ac...
Adversarial examples are one of the most intriguing topics in modern deep learning. Imperceptible... more Adversarial examples are one of the most intriguing topics in modern deep learning. Imperceptible perturbations to the input can fool robust models. In relation to this problem, attack and defense methods are being developed almost on a daily basis. In parallel, efforts are being made to simply pointing out when an input image is an adversarial example. This can help prevent potential issues, as the failure cases are easily recognizable by humans. The proposal in this work is to study how chaos theory methods can help distinguish adversarial examples from regular images. Our work is based on the assumption that deep networks behave as chaotic systems, and adversarial examples are the main manifestation of it (in the sense that a slight input variation produces a totally different output). In our experiments, we show that the Lyapunov exponents (an established measure of chaoticity), which have been recently proposed for classification of adversarial examples, are not robust to image...
Ki67 hot-spot detection and its evaluation in invasive breast cancer regions play a significant r... more Ki67 hot-spot detection and its evaluation in invasive breast cancer regions play a significant role in routine medical practice. The quantification of cellular proliferation assessed by Ki67 immunohistochemistry is an established prognostic and predictive biomarker that determines the choice of therapeutic protocols. In this paper, we present three deep learning-based approaches to automatically detect and quantify Ki67 hot-spot areas by means of the Ki67 labeling index. To this end, a dataset composed of 100 whole slide images (WSIs) belonging to 50 breast cancer cases (Ki67 and H&E WSI pairs) was used. Three methods based on CNN classification were proposed and compared to create the tumor proliferation map. The best results were obtained by applying the CNN to the mutual information acquired from the color deconvolution of both the Ki67 marker and the H&E WSIs. The overall accuracy of this approach was 95%. The agreement between the automatic Ki67 scoring and the manual analysis...
International Journal of Machine Learning and Cybernetics
Deep learning (henceforth DL) has become most powerful machine learning methodology. Under specif... more Deep learning (henceforth DL) has become most powerful machine learning methodology. Under specific circumstances recognition rates even surpass those obtained by humans. Despite this, several works have shown that deep learning produces outputs that are very far from human responses when confronted with the same task. This the case of the so-called “adversarial examples” (henceforth AE). The fact that such implausible misclassifications exist points to a fundamental difference between machine and human learning. This paper focuses on the possible causes of this intriguing phenomenon. We first argue that the error in adversarial examples is caused by high bias, i.e. by regularization that has local negative effects. This idea is supported by our experiments in which the robustness to adversarial examples is measured with respect to the level of fitting to training samples. Higher fitting was associated to higher robustness to adversarial examples. This ties the phenomenon to the tra...
Uploads
articles by Gloria Bueno
Books by Gloria Bueno
Papers by Gloria Bueno