A pipeline approach with spatial information for segmenting multiple sclerosis lesions on brain magnetic resonance imaging

Background: Conventional magnetic resonance imaging (MRI) techniques are highly sensitive to detect multiple sclerosis (MS) plaques, enabling a quantitative assessment of inflammatory activity and lesion load. In quantitative analyses of focal lesions, manual or semi-automated segmentations have been widely used to compute the total number of lesions and the total lesion volume. These techniques, however, are both challenging and time-consuming, being also prone to intra-observer and inter-observer variability.Aim: To develop an automated approach to segment brain tissues and MS lesions from brain MRI images. The goal is to reduce the user interaction and to provide an objective tool that eliminates the inter- and intra-observer variability.Methods: Based on the recent methods developed by Souplet et al. and de Boer et al., we propose a novel pipeline which includes the following steps: bias correction, skull stripping, atlas registration, tissue classification, and lesion segmentation. After the initial pre-processing steps, a MRI scan is automatically segmented into 4 classes: white matter (WM), grey matter (GM), cerebrospinal fluid (CSF) and partial volume. An expectation maximisation method which fits a multivariate Gaussian mixture model to T1-w, T2-w and PD-w images is used for this purpose. Based on the obtained tissue masks and using the estimated GM mean and variance, we apply an intensity threshold to the FLAIR image, which provides the lesion segmentation. With the aim of improving this initial result, spatial information coming from the neighbouring tissue labels is used to refine the final lesion segmentation.Results:The experimental evaluation was performed using real data sets of 1.5T and the corresponding ground truth annotations provided by expert radiologists. The following values were obtained: 64% of true positive (TP) fraction, 80% of false positive (FP) fraction, and an average surface distance of 7.89 mm. The results of our approach were quantitatively compared to our implementations of the works of Souplet et al. and de Boer et al., obtaining higher TP and lower FP values.Conclusion: Promising MS lesion segmentation results have been obtained in terms of TP. However, the high number of FP which is still a well-known problem of all the automated MS lesion segmentation approaches has to be improved in order to use them for the standard clinical practice. Our future work will focus on tackling this issue.

Eye Fundus Image Analysis for Automatic Detection of Diabetic Retinopathy

Diabetes is a rapidly increasing worldwide problem which is characterised by defective metabolism of glucose that causes long-term dysfunction and failure of various organs. The most common complication of diabetes is diabetic retinopathy (DR), which is one of the primary causes of blindness and visual impairment in adults. The rapid increase of diabetes pushes the limits of the current DR screening capabilities for which the digital imaging of the eye fundus (retinal imaging), and automatic or semi-automatic image analysis algorithms provide a potential solution. In this work, the use of colour in the detection of diabetic retinopathy is statistically studied using a supervised algorithm based on one-class classification and Gaussian mixture model estimation. The presented algorithm distinguishes a certain diabetic lesion type from all other possible objects in eye fundus images by only estimating the probability density function of that certain lesion type. For the training and ground truth estimation, the algorithm combines manual annotations of several experts for which the best practices were experimentally selected. By assessing the algorithm’s performance while conducting experiments with the colour space selection, both illuminance and colour correction, and background class information, the use of colour in the detection of diabetic retinopathy was quantitatively evaluated. Another contribution of this work is the benchmarking framework for eye fundus image analysis algorithms needed for the development of the automatic DR detection algorithms. The benchmarking framework provides guidelines on how to construct a benchmarking database that comprises true patient images, ground truth, and an evaluation protocol. The evaluation is based on the standard receiver operating characteristics analysis and it follows the medical practice in the decision making providing protocols for image- and pixel-based evaluations. During the work, two public medical image databases with ground truth were published: DIARETDB0 and DIARETDB1. The framework, DR databases and the final algorithm, are made public in the web to set the baseline results for automatic detection of diabetic retinopathy. Although deviating from the general context of the thesis, a simple and effective optic disc localisation method is presented. The optic disc localisation is discussed, since normal eye fundus structures are fundamental in the characterisation of DR.

Differentiation of Metabolically Distinct Areas within Head and Neck Region using Dynamic 18F-FDG Positron Emission Tomography Imaging

Positron Emission Tomography (PET) using ¹⁸F-FDG is playing a vital role in the diagnosis and treatment planning of cancer. However, the most widely used radiotracer, ¹⁸F-FDG, is not specific for tumours and can also accumulate in inflammatory lesions as well as normal physiologically active tissues making diagnosis and treatment planning complicated for the physicians. Malignant, inflammatory and normal tissues are known to have different pathways for glucose metabolism which could possibly be evident from different characteristics of the time activity curves from a dynamic PET acquisition protocol. Therefore, we aimed to develop new image analysis methods, for PET scans of the head and neck region, which could differentiate between inflammation, tumour and normal tissues using this functional information within these radiotracer uptake areas. We developed different dynamic features from the time activity curves of voxels in these areas and compared them with the widely used static parameter, SUV, using Gaussian Mixture Model algorithm as well as K-means algorithm in order to assess their effectiveness in discriminating metabolically different areas. Moreover, we also correlated dynamic features with other clinical metrics obtained independently of PET imaging. The results show that some of the developed features can prove to be useful in differentiating tumour tissues from inflammatory regions and some dynamic features also provide positive correlations with clinical metrics. If these proposed methods are further explored then they can prove to be useful in reducing false positive tumour detections and developing real world applications for tumour diagnosis and contouring.

Statistical segmentation methods and color variance analysis of retinal images

In this research, the effectiveness of Naive Bayes and Gaussian Mixture Models classifiers on segmenting exudates in retinal images is studied and the results are evaluated with metrics commonly used in medical imaging. Also, a color variation analysis of retinal images is carried out to find how effectively can retinal images be segmented using only the color information of the pixels.

Generative Part-Based Gabor Object Detector

Object detection is a fundamental task of computer vision that is utilized as a core part in a number of industrial and scientific applications, for example, in robotics, where objects need to be correctly detected and localized prior to being grasped and manipulated. Existing object detectors vary in (i) the amount of supervision they need for training, (ii) the type of a learning method adopted (generative or discriminative) and (iii) the amount of spatial information used in the object model (model-free, using no spatial information in the object model, or model-based, with the explicit spatial model of an object). Although some existing methods report good performance in the detection of certain objects, the results tend to be application specific and no universal method has been found that clearly outperforms all others in all areas. This work proposes a novel generative part-based object detector. The generative learning procedure of the developed method allows learning from positive examples only. The detector is based on finding semantically meaningful parts of the object (i.e. a part detector) that can provide additional information to object location, for example, pose. The object class model, i.e. the appearance of the object parts and their spatial variance, constellation, is explicitly modelled in a fully probabilistic manner. The appearance is based on bio-inspired complex-valued Gabor features that are transformed to part probabilities by an unsupervised Gaussian Mixture Model (GMM). The proposed novel randomized GMM enables learning from only a few training examples. The probabilistic spatial model of the part configurations is constructed with a mixture of 2D Gaussians. The appearance of the parts of the object is learned in an object canonical space that removes geometric variations from the part appearance model. Robustness to pose variations is achieved by object pose quantization, which is more efficient than previously used scale and orientation shifts in the Gabor feature space. Performance of the resulting generative object detector is characterized by high recall with low precision, i.e. the generative detector produces large number of false positive detections. Thus a discriminative classifier is used to prune false positive candidate detections produced by the generative detector improving its precision while keeping high recall. Using only a small number of positive examples, the developed object detector performs comparably to state-of-the-art discriminative methods.

Vidéosurveillance intelligente pour la détection de chutes chez les personnes âgées

Les pays industrialisés comme le Canada doivent faire face au vieillissement de leur population. En particulier, la majorité des personnes âgées, vivant à domicile et souvent seules, font face à des situations à risques telles que des chutes. Dans ce contexte, la vidéosurveillance est une solution innovante qui peut leur permettre de vivre normalement dans un environnement sécurisé. L’idée serait de placer un réseau de caméras dans l’appartement de la personne pour détecter automatiquement une chute. En cas de problème, un message pourrait être envoyé suivant l’urgence aux secours ou à la famille via une connexion internet sécurisée. Pour un système bas coût, nous avons limité le nombre de caméras à une seule par pièce ce qui nous a poussé à explorer les méthodes monoculaires de détection de chutes. Nous avons d’abord exploré le problème d’un point de vue 2D (image) en nous intéressant aux changements importants de la silhouette de la personne lors d’une chute. Les données d’activités normales d’une personne âgée ont été modélisées par un mélange de gaussiennes nous permettant de détecter tout événement anormal. Notre méthode a été validée à l’aide d’une vidéothèque de chutes simulées et d’activités normales réalistes. Cependant, une information 3D telle que la localisation de la personne par rapport à son environnement peut être très intéressante pour un système d’analyse de comportement. Bien qu’il soit préférable d’utiliser un système multi-caméras pour obtenir une information 3D, nous avons prouvé qu’avec une seule caméra calibrée, il était possible de localiser une personne dans son environnement grâce à sa tête. Concrêtement, la tête de la personne, modélisée par une ellipsoide, est suivie dans la séquence d’images à l’aide d’un filtre à particules. La précision de la localisation 3D de la tête a été évaluée avec une bibliothèque de séquence vidéos contenant les vraies localisations 3D obtenues par un système de capture de mouvement (Motion Capture). Un exemple d’application utilisant la trajectoire 3D de la tête est proposée dans le cadre de la détection de chutes. En conclusion, un système de vidéosurveillance pour la détection de chutes avec une seule caméra par pièce est parfaitement envisageable. Pour réduire au maximum les risques de fausses alarmes, une méthode hybride combinant des informations 2D et 3D pourrait être envisagée.

Développement d’un modèle de classification probabiliste pour la cartographie du couvert nival dans les bassins versants d’Hydro-Québec à l’aide de données de micro-ondes passives

Chaque jour, des décisions doivent être prises quant à la quantité d'hydroélectricité produite au Québec. Ces décisions reposent sur la prévision des apports en eau dans les bassins versants produite à l'aide de modèles hydrologiques. Ces modèles prennent en compte plusieurs facteurs, dont notamment la présence ou l'absence de neige au sol. Cette information est primordiale durant la fonte printanière pour anticiper les apports à venir, puisqu'entre 30 et 40% du volume de crue peut provenir de la fonte du couvert nival. Il est donc nécessaire pour les prévisionnistes de pouvoir suivre l'évolution du couvert de neige de façon quotidienne afin d'ajuster leurs prévisions selon le phénomène de fonte. Des méthodes pour cartographier la neige au sol sont actuellement utilisées à l'Institut de recherche d'Hydro-Québec (IREQ), mais elles présentent quelques lacunes. Ce mémoire a pour objectif d'utiliser des données de télédétection en micro-ondes passives (le gradient de températures de brillance en position verticale (GTV)) à l'aide d'une approche statistique afin de produire des cartes neige/non-neige et d'en quantifier l'incertitude de classification. Pour ce faire, le GTV a été utilisé afin de calculer une probabilité de neige quotidienne via les mélanges de lois normales selon la statistique bayésienne. Par la suite, ces probabilités ont été modélisées à l'aide de la régression linéaire sur les logits et des cartographies du couvert nival ont été produites. Les résultats des modèles ont été validés qualitativement et quantitativement, puis leur intégration à Hydro-Québec a été discutée.

Generative topographic mapping applied to clustering and visualization of motor unit action potentials

The identification and visualization of clusters formed by motor unit action potentials (MUAPs) is an essential step in investigations seeking to explain the control of the neuromuscular system. This work introduces the generative topographic mapping (GTM), a novel machine learning tool, for clustering of MUAPs, and also it extends the GTM technique to provide a way of visualizing MUAPs. The performance of GTM was compared to that of three other clustering methods: the self-organizing map (SOM), a Gaussian mixture model (GMM), and the neural-gas network (NGN). The results, based on the study of experimental MUAPs, showed that the rate of success of both GTM and SOM outperformed that of GMM and NGN, and also that GTM may in practice be used as a principled alternative to the SOM in the study of MUAPs. A visualization tool, which we called GTM grid, was devised for visualization of MUAPs lying in a high-dimensional space. The visualization provided by the GTM grid was compared to that obtained from principal component analysis (PCA). (c) 2005 Elsevier Ireland Ltd. All rights reserved.

A neurofuzzy classifier for two class problems

A neurofuzzy classifier identification algorithm is introduced for two class problems. The initial fuzzy base construction is based on fuzzy clustering utilizing a Gaussian mixture model (GMM) and the analysis of covariance (ANOVA) decomposition. The expectation maximization (EM) algorithm is applied to determine the parameters of the fuzzy membership functions. Then neurofuzzy model is identified via the supervised subspace orthogonal least square (OLS) algorithm. Finally a logistic regression model is applied to produce the class probability. The effectiveness of the proposed neurofuzzy classifier has been demonstrated using a real data set.

Isomap nonlinear dimensionality reduction and bimodality of Asian monsoon convection

It is known that the empirical orthogonal function method is unable to detect possible nonlinear structure in climate data. Here, isometric feature mapping (Isomap), as a tool for nonlinear dimensionality reduction, is applied to 1958–2001 ERA-40 sea-level pressure anomalies to study nonlinearity of the Asian summer monsoon intraseasonal variability. Using the leading two Isomap time series, the probability density function is shown to be bimodal. A two-dimensional bivariate Gaussian mixture model is then applied to identify the monsoon phases, the obtained regimes representing enhanced and suppressed phases, respectively. The relationship with the large-scale seasonal mean monsoon indicates that the frequency of monsoon regime occurrence is significantly perturbed in agreement with conceptual ideas, with preference for enhanced convection on intraseasonal time scales during large-scale strong monsoons. Trend analysis suggests a shift in concentration of monsoon convection, with less emphasis on South Asia and more on the East China Sea.

20th century intraseasonal Asian monsoon dynamics viewed from Isomap

The Asian summer monsoon is a high dimensional and highly nonlinear phenomenon involving considerable moisture transport towards land from the ocean, and is critical for the whole region. We have used daily ECMWF reanalysis (ERA-40) sea-level pressure (SLP) anomalies to the seasonal cycle, over the region 50-145°E, 20°S-35°N to study the nonlinearity of the Asian monsoon using Isomap. We have focused on the two-dimensional embedding of the SLP anomalies for ease of interpretation. Unlike the unimodality obtained from tests performed in empirical orthogonal function space, the probability density function, within the two-dimensional Isomap space, turns out to be bimodal. But a clustering procedure applied to the SLP data reveals support for three clusters, which are identified using a three-component bivariate Gaussian mixture model. The modes are found to appear similar to active and break phases of the monsoon over South Asia in addition to a third phase, which shows active conditions over the Western North Pacific. Using the low-level wind field anomalies the active phase over South Asia is found to be characterised by a strengthening and an eastward extension of the Somali jet whereas during the break phase the Somali jet is weakened near southern India, while the monsoon trough in northern India also weakens. Interpretation is aided using the APHRODITE gridded land precipitation product for monsoon Asia. The effect of large-scale seasonal mean monsoon and lower boundary forcing, in the form of ENSO, is also investigated and discussed. The outcome here is that ENSO is shown to perturb the intraseasonal regimes, in agreement with conceptual ideas.

Construction of neurofuzzy models for imbalanced data classification

We propose a new class of neurofuzzy construction algorithms with the aim of maximizing generalization capability specifically for imbalanced data classification problems based on leave-one-out (LOO) cross validation. The algorithms are in two stages, first an initial rule base is constructed based on estimating the Gaussian mixture model with analysis of variance decomposition from input data; the second stage carries out the joint weighted least squares parameter estimation and rule selection using orthogonal forward subspace selection (OFSS)procedure. We show how different LOO based rule selection criteria can be incorporated with OFSS, and advocate either maximizing the leave-one-out area under curve of the receiver operating characteristics, or maximizing the leave-one-out Fmeasure if the data sets exhibit imbalanced class distribution. Extensive comparative simulations illustrate the effectiveness of the proposed algorithms.

Corpus-Based Speech Enhancement With Uncertainty Modeling and Cepstral Smoothing

We present a new approach for corpus-based speech enhancement that significantly improves over a method published by Xiao and Nickel in 2010. Corpus-based enhancement systems do not merely filter an incoming noisy signal, but resynthesize its speech content via an inventory of pre-recorded clean signals. The goal of the procedure is to perceptually improve the sound of speech signals in background noise. The proposed new method modifies Xiao's method in four significant ways. Firstly, it employs a Gaussian mixture model (GMM) instead of a vector quantizer in the phoneme recognition front-end. Secondly, the state decoding of the recognition stage is supported with an uncertainty modeling technique. With the GMM and the uncertainty modeling it is possible to eliminate the need for noise dependent system training. Thirdly, the post-processing of the original method via sinusoidal modeling is replaced with a powerful cepstral smoothing operation. And lastly, due to the improvements of these modifications, it is possible to extend the operational bandwidth of the procedure from 4 kHz to 8 kHz. The performance of the proposed method was evaluated across different noise types and different signal-to-noise ratios. The new method was able to significantly outperform traditional methods, including the one by Xiao and Nickel, in terms of PESQ scores and other objective quality measures. Results of subjective CMOS tests over a smaller set of test samples support our claims.

A Pseudolikelihood Approach for Simultaneous Analysis of Array Comparative Genomic Hybridizations (aCGH)

DNA sequence copy number has been shown to be associated with cancer development and progression. Array-based Comparative Genomic Hybridization (aCGH) is a recent development that seeks to identify the copy number ratio at large numbers of markers across the genome. Due to experimental and biological variations across chromosomes and across hybridizations, current methods are limited to analyses of single chromosomes. We propose a more powerful approach that borrows strength across chromosomes and across hybridizations. We assume a Gaussian mixture model, with a hidden Markov dependence structure, and with random effects to allow for intertumoral variation, as well as intratumoral clonal variation. For ease of computation, we base estimation on a pseudolikelihood function. The method produces quantitative assessments of the likelihood of genetic alterations at each clone, along with a graphical display for simple visual interpretation. We assess the characteristics of the method through simulation studies and through analysis of a brain tumor aCGH data set. We show that the pseudolikelihood approach is superior to existing methods both in detecting small regions of copy number alteration and in accurately classifying regions of change when intratumoral clonal variation is present.

First year qualifying report: neural networks for extracting wind vectors from satellite scatterometer data

The ERS-1 Satellite was launched in July 1991 by the European Space Agency into a polar orbit at about km800, carrying a C-band scatterometer. A scatterometer measures the amount of radar back scatter generated by small ripples on the ocean surface induced by instantaneous local winds. Operational methods that extract wind vectors from satellite scatterometer data are based on the local inversion of a forward model, mapping scatterometer observations to wind vectors, by the minimisation of a cost function in the scatterometer measurement space.par This report uses mixture density networks, a principled method for modelling conditional probability density functions, to model the joint probability distribution of the wind vectors given the satellite scatterometer measurements in a single cell (the `inverse' problem). The complexity of the mapping and the structure of the conditional probability density function are investigated by varying the number of units in the hidden layer of the multi-layer perceptron and the number of kernels in the Gaussian mixture model of the mixture density network respectively. The optimal model for networks trained per trace has twenty hidden units and four kernels. Further investigation shows that models trained with incidence angle as an input have results comparable to those models trained by trace. A hybrid mixture density network that incorporates geophysical knowledge of the problem confirms other results that the conditional probability distribution is dominantly bimodal.par The wind retrieval results improve on previous work at Aston, but do not match other neural network techniques that use spatial information in the inputs, which is to be expected given the ambiguity of the inverse problem. Current work uses the local inverse model for autonomous ambiguity removal in a principled Bayesian framework. Future directions in which these models may be improved are given.