Epileptic disease is connected with change in activity of neuronal clusters. Brain connectivity analysis deals with statistic interdependencies between different neuronal centres. Earlier studies show that changes in connectivity can be seen near primary epileptic site. What is changing connectivity and its characteristic in interictal recordings are yet to be fully known. In this thesis are analyzed data from intracranial EEG electrodes, positioned in and neighboring areas of epileptic site. Changes in connectivity of epileptic site and its surroundings are observed by nonlinear correlation method. Decrease in connectivity of epileptic site during slow wave sleep was detected on frequencies above 80 Hz. Reduced connectivity was measured on the border of epileptic zone and normal tissue. Observed features are accentuated during sleep. It was also found out that connectivity at the border of epileptic zone apears to have nonlinear property. The results show that physiological processes during sleep are influencing connectivity near epileptic site and decrease in connectivity may be related to nonlinear dependence of neuronal activity at the border of epileptic zone. This study confirms hypothesis of the earlier studies and reveals new facts about connectivity of epileptic site from the perspective of nonlinear processes. Consequent study based on this findings might lead to more precise delineation of epileptic site and to better understanding of processes, which are causing epileptic fits.

This thesis describes selected methods of brain connectivity analysis. It was created an application, as a part of this thesis - plugin for evaluating relationships and dependencies between signals calculated as Pearson correlation coefficients. Computation of these coefficients is accelerated by GPU.

Data acquisition for Magnetic Resonance Imaging (MRI) is usually expensive and time-consuming. Multi-site study enables pooling more data with less cost. However, the reliability of multi-site study is not guaranteed since the data acquired from different sites always introduces site related variations. Further, these variation can not be fully resolved even using the same imaging protocols. In this thesis, we propose a seed-based image processing and statistical analyzing pipeline which mitigates the variations brought by sites to a statistically insignificant level. We collect data from a same group of subjects on two different scanners where each subject undergoes two imaging session on each site. Seed-based correlations of BOLD timeseries are used to access the connectivity between the human brain regions and seed region. The results imply that images collected from the four visits generate similar results of seed-based connectivity. The variance brought by site-related factors, machine, visit and interaction are proved to be insignificant by ANOVA test. Moreover, principle component analysis (PCA) are performed in a manner that data are reconstructed where subject identifiability is maximized. It is shown that reconstructed data introduces less variance from interaction of machine and visit.

Accurate delineation of anatomical structures in the white matter of the human brain is of paramount importance for multiple applications, such as neurosurgical planning, characterization of neurological disorders, and connectomic studies. Diffusion Magnetic Resonance Imaging (dMRI) techniques can provide, in-vivo, a mathematical representation of thousands of fibers composing such anatomical structures, in the form of 3D polylines called streamlines. Given this representation, a task of invaluable interest is known as white matter bundle segmentation, whose aim is to virtually group together streamlines sharing a similar pathway into anatomically meaningful structures, called white matter bundles. Obtaining a good and reliable bundle segmentation is however not trivial, mainly because of the intrinsic complexity of the data. Most of the current methods for bundle segmentation require extensive neuroanatomical knowledge, are time consuming, or are not able to adapt to different data settings. To overcome these limitations, the main goal of this thesis is to develop a new automatic method for accurate white matter bundle segmentation, by exploiting, combining and extending multiple up-to-date supervised learning techniques. The main contribution of the project is the development of a novel streamline-based bundle segmentation method based on binary linear classification, which simultaneously combines information from atlases, bundle geometries, and connectivity patterns. We prove that the proposed method reaches unprecedented quality of segmentation, and that is robust to a multitude of diverse settings, such as when there are differences in bundle size, tracking algorithm, and/or quality of dMRI data. In addition, we show that some of the state-of-the-art bundle segmentation methods are deeply affected by a geometrical property of the shape of the bundles to be segmented, their fractal dimension. Important factors involved in the task of streamline classification are: (i) the need for an effective streamline distance function and (ii) the definition of a proper feature space. To this end, we compare some of the most common streamline distance functions available in the literature and we provide some guidelines on their practical use for the task of supervised bundle segmentation. Moreover, we investigate the possibility to include, in a streamline-based segmentation method, additional information to the typically employed streamline distance measure. Specifically, we provide evidence that considering additional anatomical information regarding the cortical terminations of the streamlines and their proximity to specific Regions of Interest (ROIs) helps to improve the results of bundle segmentation. Lastly, significant attention is paid to reproducibility in neuroscience. Following the FAIR (Findable, Accessible, Interoperable and Reusable) Data Principles, we have integrated our pipelines of analysis into an online open platform devoted to promoting reproducibility of scientific results and to facilitating knowledge discovery.

L’autisme constitue un trouble neurodéveloppemental dont l’étiologie demeure inconnue. Les données en neuroimagerie des dernières années chez les autistes convergent vers l’observation d’une altération du transfert de l’information corticale entre les différentes régions du cerveau, plutôt qu’une atteinte sélective des structures cérébrales. Quelques modèles théoriques ont été postulés afin d’expliquer ces observations, sans toutefois unifier l’ensemble des observations. Les résultats de la littérature à ce sujet sont souvent hétérogènes et plusieurs disparités méthodologiques existent entre les études. Les conditions d’enregistrement variables et l’hétérogénéité des populations d’étude présentant souvent de multiples comorbidités limitent également leur comparaison. L’objectif de cette thèse était donc d’étudier la connectivité cérébrale de participants adultes avec autisme sans déficience intellectuelle, âgés entre 18 et 35 ans, par rapport à celle des participants neurotypiques, à l’aide d’un outil de mesure offrant une vision complémentaire à la neuroimagerie : la cohérence de l’électroencéphalographie (EEG). La cohérence de l’EEG est une méthode qui fournit de l’information quant à la synchronisation dans le temps entre paires de signaux électriques enregistrés à des sites néocorticaux distincts et constitue essentiellement une mesure de la connectivité fonctionnelle entre régions corticales. Dans cette thèse, nous avons exercé un contrôle rigoureux afin de s’assurer que nos résultats ne soient pas influencés par des variables confondantes et nous avons évalué nos participants durant le sommeil paradoxal (premier article) et lors de deux moments d’activation spontanés pendant lesquels le cortex est activé mais non sollicité, l’éveil calme yeux fermés, en soirée et au matin (deuxième article). Nous avons également évalué la relation entre les indices de cohérence significatifs à l’éveil dans le groupe avec autisme, en relation avec leurs symptômes cliniques aux questionnaires d’évaluation comportementale ADI-R et ADOS-G. Plusieurs des résultats significatifs obtenus dans cette recherche se sont avérés communs aux différents moments d’activation étudiés. En effet, l’observation d’une cohérence EEG supérieure impliquant l’aire visuelle gauche durant les états d’éveil ainsi que durant le SP semblent corroborer une certaine facilitation des régions visuelles chez les autistes par rapport au groupe contrôle. La présence d’une cohérence frontale gauche diminuée chez les participants autistes par rapport aux neurotypiques concorde avec les observations anatomiques et cliniques suggérant un déficit des fonctions cognitives impliquées dans cette région. La cohérence inter-hémisphérique frontale significativement diminuée chez les autistes par rapport aux contrôles à l’éveil du matin supporte pour sa part une altération des fibres calleuses qui pourrait être modulée par les changements développementaux associés à l’âge. Finalement, des corrélations significatives impliquant le nombre de symptômes cliniques et la cohérence EEG chez les autistes pourraient suggérer que des signes d’altération de la connectivité ont un impact sur le comportement diurne et la symptomatologie autistique. L’ensemble des résultats de cette thèse a donc permis d’approfondir les connaissances scientifiques concernant les dynamiques de connectivité cérébrale dans l’autisme et supportent l’hypothèse d’une organisation cérébrale atypique, distincte des neurotypiques, tant à l’éveil qu’au sommeil. / Autism is a neurodevelopmental disorder of unknown etiology. Converging neuroimaging data in the last years suggest that alteration in communication between regions within the autistic brain is governed by the cognitive functions associated with these regions rather than by their sheer physical distance. Some theoretical models were postulated to explain these observations, without unifying all of them. Results of the literature on this matter are often heterogeneous and several methodological disparities exist between the studies, moments and conditions of recording, and the heterogeneousness of the populations often presenting multiple comorbidity limit their interpretation. The objective of this thesis was to compare the brain connectivity of adult participants with autism (18-35 years old) without intellectual deficiency to neurotypical participants, by means of a measurement tool offering a complementary vision to the neuroimaging: the electroencephalography (EEG) coherence. The EEG coherence is a method which evaluates the synchronization in time between pairs of electrod signals recorded at different neocortical sites and constitutes essentially a measure of the functional connectivity between cortical regions. In this thesis, we exercised a rigorous control to make sure that our results are not influenced by staggering variables and we recorded our participants during REM sleep (first paper) and during two spontaneous moments of activation while the cortex is activated but not requested, waking resting state with closed eyes, during evening and morning (second paper). We also estimated the correlation between the significant EEG coherence results observed during waking state in the autism group, with their clinical symptoms on the behavioural questionnaires ADI-R and ADOS-G. Several of the significant results obtained in this research were common to all studied moments of brain activation. Indeed, the observation of a superior EEG coherence involving the left visual area during the waking states as well as during the REM sleep confirms a certain facilitation of the visual regions in the autistic group compared to the control group. The presence of a left frontal coherence decreased in the participants with autism compared to the neurotypicals supports anatomical and clinical observations suggesting a deficit of the cognitive functions involved in this region. The significantly decreased frontal inter-hemispheric coherence in the autistic group compared to the controls in the morning waking recording supports an alteration in the callosal fibers which could be modulated by developmental changes associated with age. Finally, significant correlations involving the number of clinical symptoms and the EEG coherence of autistic participants could suggest that alteration of connectivity has an impact on the diurnal behavior and the symptomatology. Thus the results of this thesis add to the scientific knowledge concerning the dynamics of cerebral connectivity in autism and support the hypothesis of an atypical brain organization, distinct from neurotypicals, both in the waking as in the sleep states.

A presente dissertação examina um método de determinação da conectividade cerebral cujo uso vem se tornando popular nos últimos anos, o partial direct coherence (PDC), que se destaca dentre outros métodos por possibilitar a verificação das relações imediatas de sinais multivariados. Este método representa a conectividade cerebral no domínio da frequência e tem íntima relação com a noção de \"causalidade\" de Granger (GRANGER, 1969), que possibilita quantificar a influência mútua entre séries temporais observadas. De um ponto de vista computacional, o referido método faz uso de modelos de séries temporais que hoje têm implementação bastante eficiente em termos de algoritmos off-line, mas cujo sucesso depende da presunção de estacionariedade dos dados, fato que é somente verdadeiro em trechos relativamente curtos de sinais de origem cerebral, como no caso do EEG (Eletroencefalograma). O objetivo deste trabalho é criar um sistema que calcule o PDC, continuamente, em tempo real e que possua a mesma precisão do método off-line, além de ser uma plataforma de estudos para implementações e testes de métodos de determinação da conectividade neural em tempo real. A plataforma desenvolvida é modular, incentivando futuros trabalhos na mesma, e mostrouse eficaz quanto a precisão numérica dos resultados do cálculo do PDC. As características de tempo real foram atingidas com algumas restrições, que dependem da configuração do usuário e do número de canais que um sinal possui. / This thesis examines a method of determination of brain connectivity whose use becomes popular in recent years, the partial direct coherence (PDC) that stands out in comparison with other methods for making possible the verification of immediate relations of multivariate signal. This method represents the brain connectivity in the frequency domain and has a close relationship with the notion of Granger causality (GRANGER, 1969) that makes it possible to quantify the mutual influence between observed time series. From a computational perspective, the above method makes use of time series models, which today has very efficient implementation in terms of off-line algorithm, but whose success depends on presume that the data is stationary, a fact that is only true in relatively short stretches of cerebral signals, especially in the case of EEG. The objective of this thesis is to create a system that calculates the PDC continuously and in real time maintaining the same precision of the off-line method. Furthermore being a research platform for implementations and tests of new methods for determining neural connectivity in real time. The developed platform is modular encouraging future work on it, and was effective in the numerical accuracy of the PDC calculation results. The real time characteristics were achieved with some restrictions that depend of the user configuration and the number of channels that the signal has.

O cérebro humano é considerado uma rede complexa em termos estruturais e funcionais em diferentes escalas. A caracterização da arquitetura desta rede pode ser considerada uma importante ferramenta no auxílio ao estudo de diferentes doenças neurodegenerativas. No presente estudo propusemos um software desenvolvido em JAVA para investigar esta arquitetura com base na correlação estatística de características morfológicas entre diferentes regiões do córtex. Foram utilizados dados de espessura cortical obtidos a partir de imagens de ressonância magnética de 191 indivíduos saudáveis e 93 pacientes com epilepsia. Foi proposto um modelo não linear para considerar o efeito da idade na espessura cortical com identificação de duas etapas: amadurecimento e envelhecimento. Os pacientes, quando comparados aos controles, apresentaram uma redução significativa na espessura cortical fundamentalmente nas regiões para-central, pericalcarina e do pré-cuneo no hemisfério direito. Esta diminuição também se manifestou ao longo da idade, com uma maior taxa de queda na região parahipocampal direita. Considerando a conectividade anatômica aqui calculada, o grupo de pacientes evidenciou alterações em 31\\% das possíveis conexões e de forma difusa. Adicionalmente, nas redes cerebrais dos pacientes houve uma diminuição de 15\\% no comprimento médio do caminho e no coeficiente de agrupamento. Aplicando-se um algoritmo de agrupamento, foram detectadas três comunidades para os indivíduos saudáveis e seis comunidades para os pacientes, confirmando uma quebra de organização estrutural neste ultimo grupo. Com este software esperamos trazer à comunidade mais uma ferramenta para análise das conexões cerebrais e suas modificações em determinadas patologias, contribuindo com seu entendimento e possível diagnóstico. / The human brain can be characterized as a complex network structurally and functionally in different levels. The description of the architecture of this network can be considered an important tool in understanding different neurodegenerative diseases. In the present study, we developed a software in JAVA to investigate this architecture based on statistical correlation of morphological characteristics between different cortex areas. It was used a database of cortical thickness obtained from magnetic resonance images of 191 healthy subjects and 93 patients with epilepsy. It was implemented a non-linear model to consider the effect of age in cortical thickness with identification of 2 stages: maturation and aging. The patients, when compared to healthy subjects, showed a significant reduction in cortical thickness, particularly at the areas precentral, pericalcarine and pré-cuneus of right hemisphere. This decrease also could be noted through the age, with a higher decrease rate at the right parahipocampal area. Considering the anatomical connectivity calculated, the patients group showed diffuse changes in 31\\% of the possible connections. Furthermore, in the patients brain network it was found a decrease of 15\\% in the characteristic path length and clustering coefficient. By applying a clustering algorithm, 3 clusters were detected in the healthy subjects and 6 clusters in the patients, confirming a breakdown of the structural organization in this last group. With our software we hope to bring to the community another tool to improve the brain connectivity analysis and its modifications in some pathologies, contributing with its understanding and possible diagnosis.

Functional magnetic resonance imaging (fMRI) is a widely used, noninvasive technique to map brain activation, and has provided considerable insight into human brain function over the past two decades. Until recently, fMRI studies have focused on gray matter; however, reports of fMRI activation in white matter are mounting. White matter fMRI activation has the potential to greatly expand the breadth of brain connectivity research, as well as improve the assessment and diagnosis of white matter and connectivity disorders. Despite these potential benefits, white matter fMRI activation remains controversial. The controversy is partially due to the existence of incompletely understood facets of fMRI signals in white matter. This thesis describes three experiments that aim to refine what is currently known about white matter fMRI activation. In the first experiment, one of the main concerns about fMRI activation in white matter was addressed; namely, whether white matter has sufficient cerebrovascular reactivity to support hemodynamic changes that can be measured with fMRI. It was demonstrated that white matter has the capacity to support detectable hemodynamic changes in the absence of partial volume effects. In the second experiment, the effect of static magnetic field strength on sensitivity to white matter fMRI activation was explored as a possible cause of the relative paucity of reports of white matter fMRI activation. The results showed greater sensitivity to white matter fMRI activation at 4 T relative to 1.5 T MRI. In the third experiment, the relationship between white matter activation and the activated network of gray matter regions was explored. This was accomplished using fMRI-guided tractography in which structural connections between activated clusters are evaluated. Structural connectivity between white matter fMRI activation and regions of gray matter activation was demonstrated, providing evidence of the functional significance of fMRI activation in white matter. These experiments provide important insights, which will allow for improved investigations of white matter fMRI activation in the future. In addition, it is posited that experimenter bias, via selective reporting of activation clusters, has contributed to the slow acceptance of fMRI activation in white matter.

A presente dissertação examina um método de determinação da conectividade cerebral cujo uso vem se tornando popular nos últimos anos, o partial direct coherence (PDC), que se destaca dentre outros métodos por possibilitar a verificação das relações imediatas de sinais multivariados. Este método representa a conectividade cerebral no domínio da frequência e tem íntima relação com a noção de \"causalidade\" de Granger (GRANGER, 1969), que possibilita quantificar a influência mútua entre séries temporais observadas. De um ponto de vista computacional, o referido método faz uso de modelos de séries temporais que hoje têm implementação bastante eficiente em termos de algoritmos off-line, mas cujo sucesso depende da presunção de estacionariedade dos dados, fato que é somente verdadeiro em trechos relativamente curtos de sinais de origem cerebral, como no caso do EEG (Eletroencefalograma). O objetivo deste trabalho é criar um sistema que calcule o PDC, continuamente, em tempo real e que possua a mesma precisão do método off-line, além de ser uma plataforma de estudos para implementações e testes de métodos de determinação da conectividade neural em tempo real. A plataforma desenvolvida é modular, incentivando futuros trabalhos na mesma, e mostrouse eficaz quanto a precisão numérica dos resultados do cálculo do PDC. As características de tempo real foram atingidas com algumas restrições, que dependem da configuração do usuário e do número de canais que um sinal possui. / This thesis examines a method of determination of brain connectivity whose use becomes popular in recent years, the partial direct coherence (PDC) that stands out in comparison with other methods for making possible the verification of immediate relations of multivariate signal. This method represents the brain connectivity in the frequency domain and has a close relationship with the notion of Granger causality (GRANGER, 1969) that makes it possible to quantify the mutual influence between observed time series. From a computational perspective, the above method makes use of time series models, which today has very efficient implementation in terms of off-line algorithm, but whose success depends on presume that the data is stationary, a fact that is only true in relatively short stretches of cerebral signals, especially in the case of EEG. The objective of this thesis is to create a system that calculates the PDC continuously and in real time maintaining the same precision of the off-line method. Furthermore being a research platform for implementations and tests of new methods for determining neural connectivity in real time. The developed platform is modular encouraging future work on it, and was effective in the numerical accuracy of the PDC calculation results. The real time characteristics were achieved with some restrictions that depend of the user configuration and the number of channels that the signal has.

Comprendre comment le cerveau humain engendre les formes les plus élaborées de comportements est profondément lié à nos connaissances générales sur son organisation anatomique et fonctionnelle. Jusqu'à récemment encore, on pensait que les fonctions cognitives n'étaient rien d'autre que le sous-produit de l'activité neurale de régions corticales discrètes et hyper-fonctionnalisées. Les découvertes majeures obtenues ces dix dernières années dans le champ de la neuro-imagerie, et plus particulièrement de la connectomique, invitent cependant à complexifier nos représentations sur les liens qu'entretiennent structures et fonctions cérébrales. Le cerveau semble en effet être organisé en systèmes neurocognitifs complexes, hautement distribués et plastiques. C'est dans cet esprit qu'a été réalisé ce travail de thèse dont l'ambition première a été de repenser les modèles actuels de la cognition sociale, et en particulier ceux ayant trait à la fonction de mentalisation, à travers l'étude comportementale des patients porteurs d'un gliome diffus de bas-grade. Cette tumeur neurologique rare constitue un excellent modèle physiopathologique en vue du démasquage des structures maîtresses des systèmes cognitifs complexes, en ce qu'elle induit des phénomènes majeurs de réorganisation fonctionnelle, et s'infiltre préférentiellement le long de la connectivité axonale associative. Des corrélations anatomo-cliniques ont été réalisées suivant une approche topologique classique (analyse de groupe en régions d'intérêt, cartographie voxel-based lesion-symptom, stimulation électrique corticale intra-opératoire) mais également hodologique (degré de déconnection des faisceaux d'association, stimulation électrique de la connectivité axonale). Les résultats principaux de nos différents travaux nous permettent de jeter les premières bases d'un modèle à double voie dynamique (plastique) et hodotopique (contraint par la réalité anatomique) de l'organisation anatomo-fonctionnelle des processus de mentalisation. Spécifiquement, une voie dorsale, interconnectant le aires corticales fronto-pariétales « miroirs » via le système périsylvien de substance blanche associative (faisceau arqué et faisceau longitudinal supérieur latéral), sous-tendrait les processus perceptifs de « bas-niveau » nécessaires à l'identification préréflexive des états mentaux ; une voie cingulo-médiane, interconnectant les régions préfrontales médiales et rostro-cingulaires aux régions pariétales postérieures médiales via le faisceau cingulaire, sous-tendrait les processus de «haut-niveau » nécessaires aux inférences mentalistiques conscientes. Ces découvertes constituent une avancée substantielle en neurosciences sociales, ont des implications importantes pour la prise en charge clinique des patients, et peuvent permettre de mieux comprendre certaines psychopathologies caractérisées à la fois par un trouble de la mentalisation et des anomalies structurales de la connectivité associative (troubles du spectre autistique). / Understanding how the brain produces sophisticated behaviours strongly depends of our knowledge on its anatomical and functional organization. Until recently, it was believed that high-level cognition was merely the by-product of the neural activity of discrete and highly specialized cortical areas. Major findings obtained in the past decade from neuroimaging, particularly from the field of connectomics, prompt now researchers to revise drastically their conceptions about the links between brain structures and functions. The brain seems indeed organized in complex, highly distributed and plastic neurocognitive networks. This is in this state of mind that our work has been carried out. Its foremost ambition was to rethink actuals models of social cognition, especially mentalizing, through the behavioural study of patients harbouring a diffuse low-grade glioma. Because this rare neurological tumour induces major functional reorganization phenomena and migrates preferentially along axonal associative connectivity, it constitutes an excellent pathophysiological model for unmasking the core structures subserving complex cognitive systems. Anatomo-clinical correlations were conducted according to both a classical topological approach (region of interest analyses, voxel-based lesion-symptom mapping, intraoperative cortical electrostimulation) and a hodological approach (degree of disconnection of associative white matter fasciculi, intraoperative axonal connectivity mapping). The main results of our different studies enable us to lay the foundation of a dynamic (plastic) and hodotopical (connectivity) dual-stream model of mentalizing. Specifically, a dorsal stream, interconnecting mirror frontoparietal areas via the perisylvian network (arcuate fasciculus and lateral superior longitudinal fasciculus), may subserve low-level perceptual processes required in rapid and pre-reflective identification of mental states; a cingulo-medial stream, interconnecting medial prefrontal and rostro-cingulated areas with medial posterior parietal areas via the cingulum, may subserve higher-level processes required in reflective mentalistic inferences. These original findings represents a great step in social neuroscience, have major implications in clinical practice, and opens new opportunities in understanding certain pathological conditions characterized by both mentalizing deficits and aberrant structural connectivity (e.g. autism spectrum disorders).