In dieser Doktorarbeit wird die Frage untersucht, wie erfolgreich deep learning bei der Diagnostik von neurodegenerativen Erkrankungen unterstützen kann. In 5 experimentellen Studien wird die Anwendung von Convolutional Neural Networks (CNNs) auf Daten der Magnetresonanztomographie (MRT) untersucht. Ein Schwerpunkt wird dabei auf die Erklärbarkeit der eigentlich intransparenten Modelle gelegt. Mit Hilfe von Methoden der erklärbaren künstlichen Intelligenz (KI) werden Heatmaps erstellt, die die Relevanz einzelner Bildbereiche für das Modell darstellen. Die 5 Studien dieser Dissertation zeigen das Potenzial von CNNs zur Krankheitserkennung auf neurologischen MRT, insbesondere bei der Kombination mit Methoden der erklärbaren KI. Mehrere Herausforderungen wurden in den Studien aufgezeigt und Lösungsansätze in den Experimenten evaluiert. Über alle Studien hinweg haben CNNs gute Klassifikationsgenauigkeiten erzielt und konnten durch den Vergleich von Heatmaps zur klinischen Literatur validiert werden. Weiterhin wurde eine neue CNN Architektur entwickelt, spezialisiert auf die räumlichen Eigenschaften von Gehirn MRT Bildern. / Deep learning and especially convolutional neural networks (CNNs) have a high potential of being implemented into clinical decision support software for tasks such as diagnosis and prediction of disease courses. This thesis has studied the application of CNNs on structural MRI data for diagnosing neurological diseases. Specifically, multiple sclerosis and Alzheimer’s disease were used as classification targets due to their high prevalence, data availability and apparent biomarkers in structural MRI data. The classification task is challenging since pathology can be highly individual and difficult for human experts to detect and due to small sample sizes, which are caused by the high acquisition cost and sensitivity of medical imaging data. A roadblock in adopting CNNs to clinical practice is their lack of interpretability. Therefore, after optimizing the machine learning models for predictive performance (e.g. balanced accuracy), we have employed explainability methods to study the reliability and validity of the trained models. The deep learning models achieved good predictive performance of over 87% balanced accuracy on all tasks and the explainability heatmaps showed coherence with known clinical biomarkers for both disorders. Explainability methods were compared quantitatively using brain atlases and shortcomings regarding their robustness were revealed. Further investigations showed clear benefits of transfer-learning and image registration on the model performance. Lastly, a new CNN layer type was introduced, which incorporates a prior on the spatial homogeneity of neuro-MRI data. CNNs excel when used on natural images which possess spatial heterogeneity, and even though MRI data and natural images share computational similarities, the composition and orientation of neuro-MRI is very distinct. The introduced patch-individual filter (PIF) layer breaks the assumption of spatial invariance of CNNs and reduces convergence time on different data sets without reducing predictive performance. The presented work highlights many challenges that CNNs for disease diagnosis face on MRI data and defines as well as tests strategies to overcome those.

No description available.

BACKGROUND: The potential pathogenesis between the presence and severity of chronic cerebrospinal venous insufficiency (CCSVI) and its relation to clinical and imaging outcomes in brain parenchyma of multiple sclerosis (MS) patients has not yet been elucidated. The aim of the study was to investigate the relationship between CCSVI, and altered brain parenchyma venous vasculature visibility (VVV) on susceptibility-weighted imaging (SWI) in patients with MS and in sex- and age-matched healthy controls (HC). METHODS: 59 MS patients, 41 relapsing-remitting and 18 secondary-progressive, and 33 HC were imaged on a 3T GE scanner using pre- and post-contrast SWI venography. The presence and severity of CCSVI was determined using extra-cranial and trans-cranial Doppler criteria. Apparent total venous volume (ATVV), venous intracranial fraction (VIF) and average distance-from-vein (DFV) were calculated for various vein mean diameter categories: < .3 mm, .3-.6 mm, .6-.9 mm and > .9 mm. RESULTS: CCSVI criteria were fulfilled in 79.7% of MS patients and 18.2% of HC (p < .0001). Patients with MS showed decreased overall ATVV, ATVV of veins with a diameter < .3 mm, and increased DFV compared to HC (all p < .0001). Subjects diagnosed with CCSVI had significantly increased DFV (p < .0001), decreased overall ATVV and ATVV of veins with a diameter < .3 mm (p < .003) compared to subjects without CCSVI. The severity of CCSVI was significantly related to decreased VVV in MS (p < .0001) on pre- and post-contrast SWI, but not in HC. CONCLUSIONS: MS patients with higher number of venous stenoses, indicative of CCSVI severity, showed significantly decreased venous vasculature in the brain parenchyma. The pathogenesis of these findings has to be further investigated, but they suggest that reduced metabolism and morphological changes of venous vasculature may be taking place in patients with MS.

Affective disorders, specifically Major Depressive Disorder and Bipolar Disorders, show high prevalence, relapse rates, and a high likelihood to develop a chronic course. For the past two decades, research has investigated the neural correlates of emotion processing and emotion regulation in patients with affective disorders. Putative underlying causal mechanisms of dysregulated affect have been informed by knowledge from the intersection of neuroimaging and clinical psychology. More recent investigations also consider processing the role of mostly negative, self-blaming social emotions, which have been linked to treatment resistance and, hence, provide a prolific target for intervention. Several psychotherapeutic treatment approaches already focus on emotion, and here specific knowledge about the mechanisms underlying persistent changes in affect bears the potential to improve the treatment of affective disorders. In this narrative review, we delineate why and how our insights into the neural correlates of emotion processing and regulation can be applied to the treatment of patients with affective disorders. / Affektive Störungen, insbesondere die Major Depression und bipolare Störungen, weisen eine hohe Prävalenz, häufige Rückfälle und eine hohe Rate an chronischen Krankheitsverläufen auf. In den letzten zwei Jahrzehnten hat die Forschung die neuronalen Korrelate der Emotionsverarbeitung und -regulation bei Patient_innen mit affektiven Störungen untersucht. Die mutmaßlichen Mechanismen der gestörten Affektregulation wurden durch Erkenntnisse aus der biologischen und klinischen Psychologie untermauert. Neuere Untersuchungen befassen sich auch mit selbstbeschuldigenden sozialen Emotionen, die mit Behandlungsresistenz in Verbindung gebracht werden und daher ein ergiebiges Ziel für Interventionen darstellen. Psychotherapeutische Behandlungsansätze konzentrieren sich bereits auf die emotionale Verarbeitung, jedoch birgt hier spezifisches Wissen über die Mechanismen, die anhaltenden affektiven Veränderungen zugrunde liegen, das Potenzial, die Behandlung von affektiven Störungen zu verbessern. In dieser narrativen Übersichtsarbeit wird dargelegt, warum und wie unsere Erkenntnisse über die neuronalen Korrelate der Emotionsverarbeitung und -regulation bei der Behandlung von Patient_innen mit affektiven Störungen eingesetzt werden können.

L’apnée obstructive du sommeil (AOS) est l’un des troubles du sommeil les plus fréquents chez l’adulte et sa prévalence augmente avec l’âge. Elle se caractérise par des arrêts répétés de la respiration au cours du sommeil, menant à la présence de fragmentation du sommeil et à de l’hypoxémie intermittente. Lorsque non traité, ce trouble peut mener à diverses conséquences non négligeables sur la santé des individus qui en sont atteints, incluant sur la santé du cerveau. L’AOS est d’ailleurs de plus en plus reconnue comme étant un possible facteur de risque de déclin cognitif et de démence. Dans ce contexte, quelques études transversales ont caractérisé le volume de la substance grise cérébrale chez des adultes vieillissants atteints d’AOS, avec des résultats variables. En effet, certaines études ont noté de plus grands volumes de substance grise chez les personnes avec une AOS plus sévère, alors que d’autres ont retrouvé des plus petits volumes chez cette même population. Ce qui explique la variabilité entre les études demeure à ce jour mal compris, bien que certaines hypothèses aient émergé. Ainsi, cette thèse vise à évaluer l’association entre la sévérité de l’AOS et le volume de substance grise cérébrale chez des personnes d’âge moyen et âgées de manière transversale et longitudinale. La première étude de cette thèse se base sur des techniques de neuroimagerie afin d’évaluer les liens entre la sévérité de l’AOS et le volume de la substance grise cérébrale des sous-régions du lobe temporal médian, soit l’hippocampe, le cortex entorhinal et le cortex parahippocampique. Celles-ci ont été ciblées puisqu’elles peuvent être affectées tôt dans la progression de la pathologie de la maladie d’Alzheimer (MA). De plus, nous avons testé l’effet d’une correction de la portion d’eau libre sur les volumes cérébraux. Finalement, nous avions comme objectif de mieux comprendre si certaines caractéristiques démographiques ou cliniques de nos participants pouvaient avoir un impact sur les associations observées. Nous avons observé qu’une AOS plus sévère était associée à des volumes de substance grise plus grands de certaines sous-régions du lobe temporal médian (hippocampe et cortex entorhinal), mais seulement chez des groupes de participants spécifiques, soit les femmes, les participants plus âgés et ceux présentant un trouble cognitif léger de type amnésique. Le fait d’apporter une correction pour la portion d’eau libre aux volumes mesurés a rendu non significatives les associations observées. Il est donc possible que la présence accrue d’eau extracellulaire, suggérant de l’œdème cérébral, puisse expliquer la présence de plus grands volumes chez les participants présentant une AOS plus sévère. La deuxième étude visait quant à elle à évaluer les changements structurels des sous-régions du lobe temporal médian associés à la sévérité de l’AOS chez des personnes d’âge moyen et âgées sur une période d’environ 2 ans. Nous avons démontré que chez nos participants n’ayant pas utilisé un traitement pour l’AOS, la présence d’interaction entre la sévérité de l’AOS et l’âge permettait d’expliquer les changements annuels de volume de substance grise. De fait, les participants plus jeunes de notre échantillon (< 65 ans) avec une AOS plus sévère présentaient un plus grand taux de changement annuel de volume de substance grise, soulignant la présence d’hypertrophie dans ce sous-groupe. Ceci a été mis en lumière pour l’ensemble des sous-régions du lobe temporal médian. Chez les participants âgés entre 65 et 75 ans, aucune association entre la sévérité de l’AOS et les changements de volume au fil du temps n’a pu être soulignée. Les participants plus âgés (> 75 ans) avec une plus grande sévérité d’AOS présentaient quant à eux une plus grande atrophie au fil du temps dans certaines régions, soit l’hippocampe et le cortex entorhinal. Ces résultats supportent donc une hypothèse biphasique des changements au niveau de la substance grise cérébrale chez les gens présentant de l’AOS, avec une première phase caractérisée par des augmentations de volume chez les adultes plus jeunes, menant éventuellement à de l’atrophie chez les personnes plus âgées. Cette thèse permet d’avoir un portrait plus clair sur la nature des changements et des mécanismes impliqués dans l’association entre la sévérité de l’AOS et les volumes de substance grise. L’un des apports importants est l’utilisation d’une nouvelle méthodologie afin d’obtenir une portion d’eau libre, ce qui a permis de mieux comprendre l’apport potentiel de mécanismes pouvant sous-tendre les changements structuraux observés, notamment l’œdème cérébral. De plus, l’évaluation des caractéristiques individuelles des participants a permis d’expliquer partiellement les incongruences entre les études précédentes. Dans le cadre des études incluses dans cette thèse, nous avons observé des changements plus marqués chez les femmes. Nous avons également pu démontrer que l’âge des individus atteints d’AOS pouvait influencer significativement le patron de changements observés. Les résultats de cette thèse pourraient donc permettre de mieux cibler les personnes avec AOS qui pourraient le plus bénéficier d’un traitement pour maintenir leur santé cérébrale. / Obstructive sleep apnea (OSA) is one of the most common sleep disorders in adults, and its prevalence increases with age. It is characterized by repeated pauses in breathing during sleep, leading to sleep fragmentation and intermittent hypoxemia. If left untreated, this disorder can have numerous consequences, including on the brain’s health. OSA is increasingly recognized as a risk factor for cognitive decline and dementia. In this context, cross-sectional studies have characterized brain gray matter volume in aging adults with OSA, with variable results. Indeed, some studies have noted greater gray matter volumes in people with more severe OSA, while others have found smaller volumes in this same population. What explains the variability between studies remains poorly understood, although some hypotheses have emerged. Thus, this thesis aims to assess the association between OSA severity and cerebral gray matter volume in middle-aged and elderly individuals using cross-sectional and longitudinal designs. The first study in this thesis uses neuroimaging techniques to assess the links between OSA severity and cerebral gray matter volume of the medial temporal lobe subregions, i.e. the hippocampus, entorhinal cortex and parahippocampal cortex. These were chosen as they can be affected early in the progression of Alzheimer's disease (AD) pathology. We also corrected our brain volumes for free-water portion. Finally, we aimed to better understand whether certain demographic or clinical characteristics of our participants might have an impact on the associations observed. We noted that more severe OSA was associated with larger gray matter volumes in certain subregions of the medial temporal lobe (hippocampus and enthorinal cortex), but only in specific groups of participants: women, older participants and those with amnestic mild cognitive impairment. Correcting our volumes for free-water portion rendered the associations nonsignificant. It is therefore possible that the presence of extracellular water, suggestive of cerebral edema, could explain the presence of larger volumes in participants with more severe OSA. The second study aimed to assess longitudinal structural changes associated with OSA severity in middle-aged and elderly people over a period of around 2 years. We found that in participants who did not use treatment for OSA, the presence of interactions between OSA severity and age were associated with the annual changes in gray matter volume. Indeed, younger participants (< 65 years old) in our sample with more severe OSA showed a greater rate of annual change in gray matter volume, highlighting the presence of hypertrophy in this subgroup. This was underlined in all medial temporal lobe subregions. In participants aged between 65 and 75, no association between OSA severity and volume changes over time could be highlighted. Older participants (>75 years old) with greater OSA severity showed greater hippocampal and entorhinal cortex atrophy over time. These results therefore support a biphasic hypothesis of changes in cerebral gray matter in people with OSA, with an initial phase characterized by volume increases in younger adults, eventually leading to atrophy in older people. This thesis provides a clearer picture of the nature of the changes and mechanisms involved in the association between OSA severity and gray matter volumes. An important contribution is the use of a new methodology to obtain a free-water portion, which allows to better understand the potential contribution of mechanisms that may underlie the structural changes observed, notably cerebral edema. In addition, the assessment of participants' individual characteristics helped to partially explain incongruities between previous studies. Indeed, in the studies included in this thesis, we observed more marked changes in certain subgroups of participants, notably women. We were also able to demonstrate that the age of individuals with OSA could significantly influence the pattern of changes observed, either gray matter hypertrophy or atrophy. The results of this thesis could therefore make it possible to target specific subgroups of individuals suffering from OSA who may be at greater risk of displaying changes in gray matter structure, and thus promote screening and treatment when necessary.

Infants with hypoxic-ischaemic encephalopathy (birth asphyxia) have a high risk of death or disability. Those with poor prognosis are sometimes allowed to die after withdrawal of intensive care. In recent years, doctors have used new types of brain scan, magnetic resonance imaging (MRI), to predict the type and severity of impairment if the infant survives and to help with such decisions. In this thesis, I analyse the issues arising from the use of MRI for prognostication and decision-making in newborn infants. I argue that previous prognostic research has been hampered by a failure to identify and focus on the most important practical question and that this contributes to uncertainty in practice. I outline recommendations for improving research. I then look at existing guidelines about withdrawal of life-sustaining treatment. I identify several problems with these guidelines; they are vague and fail to provide practical guidance, they provide little or no genuine scope for parental involvement in decisions, and they give no weight to the interests of others. I argue that parental interests should be given some weight in decisions for newborn infants. I develop a new model of decision-making that, using the concept of a Restricted Life, attempts to set out clearly the boundaries of parental discretion in decision-making. I argue that where infants are predicted to have severe cognitive or very severe physical impairment parents should be permitted to request either withdrawal or continuation of treatment. I justify this model on the basis of overlapping interests, (prognostic, experiential and moral) uncertainty, asymmetrical harms, and the burden of care. In the conclusion, I set out a guideline for the use of MRI in newborn infants with hypoxic-ischaemic encephalopathy. I suggest that this guideline would provide a more robust, coherent and practical basis for decision-making in newborn intensive care.

A statistical meta-analysis combines the results of several studies that address a set of related research hypotheses, thus increasing the power and reliability of the inference. Meta-analytic methods are over 50 years old and play an important role in science; pooling evidence from many trials to provide answers that any one trial would have insufficient samples to address. On the other hand, the number of neuroimaging studies is growing dramatically, with many of these publications containing conflicting results, or being based on only a small number of subjects. Hence there has been increasing interest in using meta-analysis methods to find consistent results for a specific functional task, or for predicting the results of a study that has not been performed directly. Current state of neuroimaging meta-analysis is limited to coordinate-based meta-analysis (CBMA), i.e., using only the coordinates of activation peaks that are reported by a group of studies, in order to "localize" the brain regions that respond to a certain type of stimulus. This class of meta-analysis suffers from a series of problems and hence cannot result in as accurate results as desired. In this research, we describe the problems that existing CBMA methods are suffering from and introduce a hierarchical mixed-effects image-based metaanalysis (IBMA) solution that incorporates the sufficient statistics (i.e., voxel-wise effect size and its associated uncertainty) from each study. In order to improve the statistical-inference stage of our proposed IBMA method, we introduce a nonparametric technique that is capable of adjusting such an inference for spatial nonstationarity. Given that in common practice, neuroimaging studies rarely provide the full image data, in an attempt to improve the existing CBMA techniques we introduce a fully automatic model-based approach that employs Gaussian-process regression (GPR) for estimating the meta-analytic statistic image from its corresponding sparse and noisy observations (i.e., the collected foci). To conclude, we introduce a new way to approach neuroimaging meta-analysis that enables the analysis to result in information such as “functional connectivity” and networks of the brain regions’ interactions, rather than just localizing the functions.

Introdução: A esclerodermia é rara doença do tecido conectivo que se manifesta através da esclerose cutânea e variável acometimento sistêmico. Duas categorias de esclerodermia são conhecidas: esclerose sistêmica, caracterizada por esclerose cutânea e acometimento visceral e a esclerodermia localizada ou morfeia que classicamente apresenta evolução benigna e autolimitada, confinada a pele e/ou tecidos subjacentes. Estudos recentes demonstram que a forma localizada possa eventualmente apresentar acometimento de órgãos internos e morbidade variável. Objetivo: Neste estudo objetivamos determinar as características demográficas, a prevalência de manifestações sistêmicas e alterações laboratoriais, bem como a associação com doenças autoimunes, em pacientes com esclerodermia da face. Métodos: Pacientes com esclerodermia localizada, incluindo os casos de esclerodermia em golpe de sabre, síndrome de Parry-Romberg e morfeia em placas com acometimento facial, foram avaliados e submetidos à investigação neurológica, incluindo exame clínico neurológico e ressonância magnética de crânio, e avaliação oftalmológica. Após 3 anos, o subgrupo de pacientes disponível para seguimento foi ressubmetido à ressonância magnética. Resultados: Foram estudados 34 pacientes com esclerodermia localizada da face. Deste total, 64,7% apresentavam uma ou mais manifestações extracutâneas, sendo cefaleia a queixa mais frequente, encontrada em 61,8% dos pacientes. Dos 23 pacientes submetidos à avaliação neurológica, 56,5% apresentaram alterações neurológicas possivelmente associadas à esclerodermia. Alterações à ressonância magnética foram observadas em 50% dos casos. O achado mais frequente foi a presença de lesões parenquimatosas com alteração de sinal em 50% dos pacientes. Dos pacientes que apresentavam alteração neurológica, 80% também apresentavam alguma alteração à ressonância magnética. Doze pacientes foram ressubmetidos a novo exame após 3 anos. Em todos os pacientes os achados de imagem se mantiveram inalterados. Durante esse intervalo de 3 anos, 25% dos pacientes apresentaram sinais de atividade da esclerodermia. Quanto à avaliação oftalmológica, 67,9% dos pacientes avaliados apresentaram alteração, sendo os achados mais frequentes a ocorrência de alterações orbiculares da esclerodermia (20,6%) e xeroftalmia (10,7%). Conclusão: Pacientes com esclerodermia localizada da face apresentam alta prevalência de alterações neurológicas e oftalmológicas. Baseado nestes achados, sugerimos que todos os casos de esclerodermia localizada da face devam ser detalhadamente examinados quanto à presença de alterações sistêmicas / Introduction: Scleroderma is a rare connective tissue disease that manifests as skin sclerosis and variable systemic involvement. Two categories of scleroderma are known: systemic sclerosis, characterized by cutaneous sclerosis and visceral involvement and localized scleroderma or morphea which classically presents benign evolution and selflimited, confined to the skin and / or underlying tissue. Recent studies show that the localized form may possibly course with involvement of internal organs and variable morbidity. Objective: This study aimed to determine the demographic characteristics, the prevalence of systemic manifestations and laboratory findings, as well as the association with autoimmune diseases, in patients with scleroderma of the face. Methods: Patients with localized scleroderma, including cases of scleroderma en coup de sabre, Parry-Romberg syndrome and morphea plaque with facial involvement were evaluated and underwent neurological examination, including neurologic examination and magnetic resonance imaging, and ophthalmology evaluation. After 3 years, the subgroup of patients available for follow-up was subjected again to MRI. Results: We studied 34 patients with localized scleroderma of the face. Of this total, 64,7% had one or more extracutaneous manifestation, headache being the most frequent complaint found in 61,8% of patients. Of the 23 patients undergoing neurological evaluation, 56,5% had neurological changes possibly associated with scleroderma. MRI changes were observed in 50% of cases. The most frequent was the presence of parenchymal lesions with signal alteration in 50% of patients. Of the patients who had neurological deficits, 80% also had a change to MRI. Twelve patients were subjected again to another MRI scan after 3 years. In all patients, imaging findings were unchanged. During this interval of 3 years, 25% of patients showed signs of activity of scleroderma. As for ophthalmologic evaluation, 67,9% of patients showed abnormalities, with the most frequent findings being the occurrence of orbicular changes of scleroderma (20.6%) and xerophthalmia (10.7%). Conclusion: Patients with localized scleroderma face have a high prevalence of neurological and ophthalmological changes. Based on these findings, we suggest that all cases of localized scleroderma of the face should be thoroughly examined for the presence of systemic changes

Environ 150.000 personnes souffrent en France d'une épilepsie partielle réfractaire à tous les médicaments. La chirurgie, qui constitue aujourd’hui le meilleur recours thérapeutique nécessite un bilan préopératoire complexe. L'analyse de données d'imagerie telles que l’imagerie par résonance magnétique (IRM) anatomique et la tomographie d’émission de positons (TEP) au FDG (fluorodéoxyglucose) tend à prendre une place croissante dans ce protocole, et pourrait à terme limiter de recourir à l’électroencéphalographie intracérébrale (SEEG), procédure très invasive mais qui constitue encore la technique de référence. Pour assister les cliniciens dans leur tâche diagnostique, nous avons développé un système d'aide au diagnostic (CAD) reposant sur l'analyse multivariée de données d'imagerie. Compte tenu de la difficulté relative à la constitution de bases de données annotées et équilibrées entre classes, notre première contribution a été de placer l'étude dans le cadre méthodologique de la détection du changement. L'algorithme du séparateur à vaste marge adapté à ce cadre là (OC-SVM) a été utilisé pour apprendre, à partir de cartes multi-paramétriques extraites d'IRM T1 de sujets normaux, un modèle prédictif caractérisant la normalité à l'échelle du voxel. Le modèle permet ensuite de faire ressortir, dans les images de patients, les zones cérébrales suspectes s'écartant de cette normalité. Les performances du système ont été évaluées sur des lésions simulées ainsi que sur une base de données de patients. Trois extensions ont ensuite été proposées. D'abord un nouveau schéma de détection plus robuste à la présence de bruit d'étiquetage dans la base de données d'apprentissage. Ensuite, une stratégie de fusion optimale permettant la combinaison de plusieurs classifieurs OC-SVM associés chacun à une séquence IRM. Enfin, une généralisation de l'algorithme de détection d'anomalies permettant la conversion de la sortie du CAD en probabilité, offrant ainsi une meilleure interprétation de la sortie du système et son intégration dans le bilan pré-opératoire global. / One third of patients suffering from epilepsy are resistant to medication. For these patients, surgical removal of the epileptogenic zone offers the possibility of a cure. Surgery success relies heavily on the accurate localization of the epileptogenic zone. The analysis of neuroimaging data such as magnetic resonance imaging (MRI) and positron emission tomography (PET) is increasingly used in the pre-surgical work-up of patients and may offer an alternative to the invasive reference of Stereo-electro-encephalo -graphy (SEEG) monitoring. To assist clinicians in screening these lesions, we developed a computer aided diagnosis system (CAD) based on a multivariate data analysis approach. Our first contribution was to formulate the problem of epileptogenic lesion detection as an outlier detection problem. The main motivation for this formulation was to avoid the dependence on labelled data and the class imbalance inherent to this detection task. The proposed system builds upon the one class support vector machines (OC-SVM) classifier. OC-SVM was trained using features extracted from MRI scans of healthy control subjects, allowing a voxelwise assessment of the deviation of a test subject pattern from the learned patterns. System performance was evaluated using realistic simulations of challenging detection tasks as well as clinical data of patients with intractable epilepsy. The outlier detection framework was further extended to take into account the specificities of neuroimaging data and the detection task at hand. We first proposed a reformulation of the support vector data description (SVDD) method to deal with the presence of uncertain observations in the training data. Second, to handle the multi-parametric nature of neuroimaging data, we proposed an optimal fusion approach for combining multiple base one-class classifiers. Finally, to help with score interpretation, threshold selection and score combination, we proposed to transform the score outputs of the outlier detection algorithm into well calibrated probabilities.

O transtorno obsessivo-compulsivo (TOC) é um transtorno psiquiátrico que acomete cerca de 1 a 3,1% das pessoas ao longo da vida. Embora o seu modelo neurobiológico ainda não esteja completamente estabelecido, inúmeras evidências apontam para áreas relacionadas ao circuito córtico-estriado-pálido-talâmico-cortical (CEPTC). Em especial, o córtex órbito-frontal (COF) é uma região que desempenha um papel fundamental dentro da hipótese fisiopatológica do TOC. Paralelamente, esta região já foi associada, em sujeitos saudáveis, com a habilidade de utilização espontânea da estratégia de agrupamento semântico na memorização de palavras - o que facilita sua evocação posterior. Ao mesmo tempo, estudos neuropsicológicos evidenciaram que pacientes com TOC apresentam déficits na memória episódica verbal (MEV) e que tais déficits poderiam ser mediados por dificuldades em funções executivas ligadas ao planejamento, como utilização de estratégias. Portanto, para testar a hipótese de que há diferenças no correlato neural da codificação da MEV entre pacientes com TOC e controles saudáveis, foi utilizado um teste neuropsicológico adaptado para ressonância magnética funcional (RMf): o paradigma tinha apresentação em bloco. O objetivo do presente estudo foi investigar a etapa de codificação da MEV e a capacidade de agrupamento semântico espontâneo em crianças e adolescentes com TOC. Assim, o paradigma foi constituído por duas listas de palavras: uma, semanticamente relacionada (SR), na qual as palavras eram divididas em categorias semânticas e outra, não relacionada (NR), na qual não havia relação aparente entre as palavras. O contraste de maior interesse do estudo foi a diferença entre essas duas condições (SR > NR). O nível de agrupamento semântico foi quantificado por um índice semântico. Os grupos foram formados por 25 crianças e adolescentes com TOC e 25 controles saudáveis, pareados por sexo, idade, escolaridade, preferência manual e QI. Embora os grupos estivessem pareados por essas características, eles se diferiram em sintomas clínicos, tais como sintomas de depressão, ansiedade e necessidade de rotina por parte da criança/adolescente. Os resultados comportamentais do teste de MEV mostraram que os grupos não se diferenciaram: ambos evocaram a mesma quantidade de palavras e não apresentaram diferenças no índice semântico. Apesar disso, a comparação entre os grupos - controlada para variáveis clínicas - revelou menor ativação (sinal BOLD) nos pacientes em diversas regiões cerebrais: frontais, parietais e occipito-temporais. Por outro lado, a análise de interação psicofisiológica (PPI) revelou que os pacientes apresentaram um aumento da conectividade do COF com regiões temporais em relação aos controles. Isso ocorreu para três das quatro regiões de interesse que foram posicionadas no COF: lateral e medial de ambos os hemisférios. Além disso, o grupo de pacientes apresentou uma correlação positiva entre o índice semântico e o efeito BOLD no COF, o que não ocorreu para o grupo controle. Esses resultados indicam diferenças no funcionamento cerebral de crianças e adolescentes com TOC tanto em regiões que estão dentro do modelo neurobiológico proposto para o TOC (circuito CEPTC), como fora dele também. De acordo com os resultados do presente estudo, as diferenças de ativação e de conectividade poderiam ser consideradas como um déficit latente, uma vez que ambos os grupos apresentaram o mesmo desempenho no paradigma / The obsessive-compulsive disorder (OCD) is a psychiatric disorder that affects 1-3.1% of the general population (lifetime rate). Although its neurobiological model has not been completely establish, numerous evidences indicate that areas of the cortico-striatalpale- thalamic-cortical (CSPTC) circuit are engaged in the disease. In particular, the orbitofrontal cortex (OFC) is a region that plays a key role in the pathophysiological hypothesis of OCD. In parallel to this, in healthy controls this region has been associated with the ability of using spontaneous strategies of semantic clustering at the encoding of related words - in a way that facilitates the posterior retrieval of these words. At the same time, neuropsychological studies showed that OCD patients present verbal episodic memory (VEM) deficits, and that these deficits could be mediated by executive dysfunction - like planing and utilization of strategies. Thus, to investigate the hypothesis that there are differences at the neural correlates of VEM encoding between children and adolescents with OCD and healthy controls, we used a blocked design functional Magnetic Resonance Imaging (fMRI) paradigm to evaluate both groups. The main objective of the study was to investigate the VEM encoding and the ability to spontaneously organize words according to their semantic categories. In order to do this, the fMRI paradigm consisted of two kinds of word lists: a semantically related list (SR), in which words were divided into semantic categories and a unrelated list (UR), were there was no apparent relationship between the words. However, the contrast of most interest of this study, was the difference between the conditions (\'SR > UR\'). The semantic clustering level was quantified by a semantic clustering index. Groups were constituted by 25 children and adolescents with OCD and 25 healthy controls paired by gender, age, educational level, handedness and IQ. Although both groups were matched for these characteristics, they differed in clinical symptoms such as depression, anxiety and routines. Behavioral results showed that the groups were similar in terms of retrieved words and semantic index. Nevertheless, the comparison between groups - controlled for clinical variables - showed less activation (BOLD signal) in patients in several brain regions: frontal, parietal and occipito-temporal. On the other hand, the psychophysiological interaction analysis (PPI) revealed that patients have had an increase in the OFC connectivity with the temporal regions. This has occurred in three of the four regions of interest that were placed in the OFC: lateral and medial of both hemispheres. Also, the patients showed a positive correlation between the semantic index and the BOLD effect in the OFC, which was not observed in the control group. These results suggest that there are differences in brain functioning of children and adolescents with OCD in regions that are inside/outside of the neurobiological model for OCD (CSPTC circuit). In accordance with the present results, these differences in brain activation and connectivity could be regarded as a latent deficit, since both groups presented the same behavioral performance