L'IRM de diffusion (IRMd) est une modalité d'imagerie médicale qui suscite un intérêt croissant dans la recherche en neuro-imagerie. Elle permet d'apporter in vivo des informations nouvelles sur les micro-structures locales des tissus. En chaque point d'une acquisition d'IRMd, la distribution des directions de diffusion des molécules d'eau est modélisée par un tenseur de diffusion. La nature multivariée de ces images requiert la conception de nouvelles méthodes de traitement adaptées. Le contexte de cette thèse est l'analyse automatique de changements longitudinaux intra-patient avec pour application le suivi de pathologies neuro-dégénératives. Notre recherche a ainsi porté sur le développement de nouveaux modèles et tests statistiques permettant la détection automatique de changements sur des séquences temporelles d'images de diffusion. Cette thèse a ainsi permis une meilleure prise en compte de la nature tensorielle des modèles d'ordre 2 (tests statistiques sur des matrices définies positives), une extension vers des modèles d'ordre supérieur et une gestion plus fine des voisinages sur lesquels les tests sont menés, avec en particulier la conception de tests statistiques sur des faisceaux de fibres. / Diffusion MRI is a new medical imaging modality of great interest in neuroimaging research. This modality enables the characterization in vivo of local micro-structures. Tensors have commonly been used to model the diffusivity profile at each voxel. This multivariate data set requires the design of new dedicated image processing techniques. The context of this thesis is the automatic analysis of intra-patient longitudinal changes with application to the follow-up of neuro-degenerative pathologies. Our research focused on the development of new models and statistical tests for the automatic detection of changes in temporal sequences of diffusion images. Thereby, this thesis led to a better modeling of second order tensors (statistical tests on positive definite matrices), to an extension to higher-order models, and to the definition of refined neighborhoods on which tests are conducted, in particular the design of statistical tests on fiber bundles.

L’imagerie par résonance magnétique pondérée en diffusion est une modalité unique sensible aux mouvements microscopiques des molécules d’eau dans les tissus biologiques. Il est possible d’utiliser les caractéristiques de ce mouvement pour inférer la structure macroscopique des faisceaux de la matière blanche du cerveau. La technique, appelée tractographie, est devenue l’outil de choix pour étudier cette structure de façon non invasive. Par exemple, la tractographie est utilisée en planification neurochirurgicale et pour le suivi du développement de maladies neurodégénératives.Dans cette thèse, nous exposons certains des biais introduits lors de reconstructions par tractographie, et des méthodes sont proposées pour les réduire. D’abord, nous utilisons des connaissances anatomiques a priori pour orienter la reconstruction. Ainsi, nous montrons que l’information anatomique sur la nature des tissus permet d'estimer des faisceaux anatomiquement plausibles et de réduire les biais dans l’estimation de structures complexes de la matière blanche. Ensuite, nous utilisons des connnaissances microstructurelles a priori dans la reconstruction, afin de permettre à la tractographie de suivre le mouvement des molécules d’eau non seulement le long des faisceaux, mais aussi dans des milieux microstructurels spécifiques. La tractographie peut ainsi distinguer différents faisceaux, réduire les erreurs de reconstruction et permettre l’étude de la microstructure le long de la matière blanche. Somme toute, nous montrons que l’utilisation de connaissances anatomiques et microstructurelles a priori, en tractographie, augmente l’exactitude des reconstructions de la matière blanche du cerveau. / Diffusion-weighted magnetic resonance imaging is a unique imaging modality sensitive to the microscopic movement of water molecules in biological tissues. By characterizing the movement of water molecules, it is possible to infer the macroscopic neuronal pathways of the brain. The technique, so-called tractography, had become the tool of choice to study non-invasively the human brain's white matter in vivo. For instance, it has been used in neurosurgical intervention planning and in neurodegenerative diseases monitoring. In this thesis, we report biases from current tractography reconstruction and suggest methods to reduce them. We first use anatomical priors, derived from a high resolution T1-weighted image, to guide tractography. We show that knowledge of the nature of biological tissue helps tractography to reconstruct anatomically valid neuronal pathways, and reduces biases in the estimation of complex white matter regions. We then use microstructural priors, derived from the state-of-the-art diffusionweighted magnetic resonance imaging protocol, in the tractography reconstruction process. This allows tractography to follow the movement of water molecules not only along neuronal pathways, but also in a microstructurally specific environment. Thus, the tractography distinguishes more accurately neuronal pathways and reduces reconstruction errors. Moreover, it provides the mean to study white matter microstructure characteristics along neuronal pathways. Altogether, we show that anatomical and microstructural priors used during the tractography process improve brain’s white matter reconstruction.

Gray matter (GM) forms the ‘computational engine’ of our brain and plays the key role in brain function. Measures derived from MRI (e.g., structural MRI (sMRI) and diffusion MRI (dMRI)) provide a unique opportunity to non-invasively study GM structure in-vivo and thus can be used to probe GM pathology in development, aging and neuropsychiatric disorders. Investigation of the influence of various factors on MRI measures in GM is critical to facilitate their use for future non-invasive studies in healthy and diseased populations. In this dissertation, GM structure was studied with MRI to understand how it is influenced by genetic and environmental factors. Validation of dMRI- derived measures was conducted by comparing them with histological data from monkeys to better understand the cytoarchitectural features that influence GM measures. First, the influence of genetic and environmental factors was quantified on gray matter macrostructure and microstructure measures using phenotypic modelling of structural and diffusion MRI data obtained from a large twin and sibling population (N = 840). Results of this study showed that in GM, while sMRI measures like cortical thickness and GM volume are mainly affected by genetic factors, advanced dMRI measures of mean squared displacement (MSD) and return to origin probability (RTOP) derived from advanced biexponential model can tap into regionally specific patterns of both genetic and environmental influence in cortical and subcortical GM. Our results thus highlight the potential of these advanced dMRI measures for use in future studies that aim to investigate and follow in healthy and clinical population changes in GM microstructure linked with both genes and environment. Second, using data from a large healthy population (n=550), we investigated changes in sMRI tissue contrast at the gray-white matter boundary with biological development during adolescence to assess how this affects estimation of the developmental trajectory of cortical thickness. Results of this study suggest that increased myelination during brain development contributes to age-related changes in gray-white boundary contrast in sMRI scans causing an apparent shift of the estimated gray-white boundary towards the cortical surface, in turn reducing estimations of cortical thickness and its developmental trajectory. Based on these results, we emphasize the importance of accounting for the effects of myelination on T1 gray-white matter boundary contrast to enable more precise estimation of cortical thickness during neurodevelopment. Finally, we conducted histological validation of dMRI measures in gray matter by comparing dMRI measures derived from two models, conventional Diffusion Tensor Imaging (DTI) model and an advanced biexponential model with histology acquired from the same 4 rhesus monkeys. Results demonstrate differences in the ability of distinct dMRI measures including DTI-derived measures of fractional anisotropy (FA), Trace and advanced Biexponential model-derived measures of MSD and RTOP to capture the biological features of underlying cytoarchitecture and identify the dMRI measures that best reflect underlying gray matter cytoarchitectural properties. Investigation of the contribution of underlying cytoarchitecture (cellular organization) to dMRI measures in gray matter provides validation of dMRI measures of average and regional heterogeneity in MSD & Trace as markers of cytoarchitecture as measured by regional average and heterogeneity in cell area density. This postmortem validation of these dMRI measures makes their use possible for treatment monitoring of various GM pathologies. These studies and their results together demonstrate the utility of imaging measures to investigate the complex relationships between GM cellular organization, brain development, environment and genes.

Diffusion Magnetic Resonance Imaging (DMRI) is a class of Magnetic Resonance Imaging (MRI) techniques with broad medical applications ranging from characterization of tumors and brain damage to potential prediction of stroke. Gradient coil and signal-to- noise ratio (SNR) constraints limit spatial resolution, accuracy, and scan time in DMRI. Achieving high b-values (measures of a scan's sensitivity to diffusion) often require scans with long diffusion gradient pulses, leading to significant magnetic resonance (MR) signal decay before the signal can be sampled. This signal loss reduces the accuracy of diffusion parameter estimation. The ability to sample the MR signal sooner while maintaining the same b-value is restricted by the maximum amplitude and slew rate of gradient coils. A composite system utilizing body and high-powered insert gradient coils can achieve high b-values more quickly, enabling a shorter delay between excitation and signal sampling and improved accuracy of diffusion parameter estimation. Alternately, such a system can achieve higher b-values at an equivalent delay between excitation and signal sampling. This thesis describes the implementation of such a system, experiment design for evaluating the benefits of the system to DMRI, and design of a diffusion phantom. Also included are a characterization of a composite system's improvements to DMRI based on analysis of experimentally-obtained data and simulation results validating those findings. Finally, recommendations for further improvements to diffusion MRI are given.

Measuring transmembrane water exchange can provide potential biomarkers for tumors and brain disorders. Diffusion Magnetic Resonance Imaging (dMRI) is a well-established tool that can non-invasively measure water exchange across cell membranes. Diffusion Exchange Spectroscopy (DEXSY) is one of the dMRI-based frameworks used to estimate exchange. DEXSY provides a detailed picture of multi-site exchange processes but requires a large quantity of data. Several models based on the DEXSY framework have been proposed to reduce the acquisition time. Filter Exchange Imaging (FEXI) and curvature models are two of them that only require certain samples of the DEXSY dataset. Diffusion-Exchange Weighted (DEW) Imaging model is another data reduction method accounting for restricted diffusion within cells and can use a specific subset of the DEXSY dataset to measure exchange. Furthermore, a more general expression of the DEXSY signal, referred to as the general model, can theoretically analyze the full space or reduced DEXSY datasets and estimate exchange. However, the results of the subsampling schemes and the data reduction models have not been compared to the full space estimation.  Therefore, this thesis aims to experimentally explore the feasibility of estimating exchange using these four models (the general, FEXI, curvature and DEW models) with the data acquired using a low-field benchtop MR scanner, and compare the estimates from the general model with different subsampling schemes and the data reduction models to the full space estimation. For this purpose, a double diffusion encoding (DDE) sequence was modified from an existing sequence on the benchtop MR scanner and a DEXSY experiment was conducted on this MR scanner and a yeast phantom to acquire a full space dataset. The exchange parameters estimated from the full space dataset using the general model were used as "ground truths" to evaluate the estimates from the reduced datasets analyzed using the general, FEXI and curvature models. Moreover, two alternative subsampling schemes named the shifted DEW and new trajectory schemes were proposed and employed to measure exchange. The results indicate that all the methods except the curvature sampling scheme employed with both the general and curvature models provided comparable estimates to the "ground truths". The shifted DEW and new trajectory sampling schemes performed better over others in terms of consistency with the "ground truths" and low variations between voxels, suggesting the theoretical and experimental optimization of these two subsampling schemes can be further studied and developed.

Diffusion-weighted magnetic resonance imaging (dMRI) is a powerful tool for thecharacterisation of neural tissue microstructural features. The role of neural projectioncurvature on the diffusion signal was recently studied for three temporal regimes of the diffusion pulse sequence in search for a description of the different decay trends in the orientationally-averaged diffusion signal reported in in vivo human studies.This work experimentally investigates the effects of neural projection curvedness in one of these regimes, namely the short diffusion time regime. Multi-shell diffusion MRI acquisitions on fixed rat spinal cord were performed using a custom number of diffusion gradient directions on a vertical bore pre-clinical MRI scanner capable of generating 3000 mT/m. Diffusion was probed in three different q-values ranges [450, 970], [600, 1400] and [1500, 1750] mm-1 using diffusion pulse durations of 1.4,2 and 2.5ms, respectively. Noise correction was performed on the diffusion data and the orientationally-averaged signal was computed for each shell using a weighted mean. The signal from selected regions in the sample was then fitted to a power law. Results show that gray matter areas exhibit a signal reduction with variable decay trends in the range of diffusion sensitivity values used here. This suggests that gray matter microstructure features are pictured by the orientationally-averaged signal in the high diffusion sensitivity regime and, as theoretically suggested, neurite curvature might play a role in characterizing the signal decay. These preliminary results may prove useful in the development of models for the interpretation of the diffusion signal and the design of acquisition strategies that aim to study the high diffusion sensitivity regime.

La magnéto-encéphalographie (MEG) mesure l´activité cérébrale avec un excellent décours temporel mais sa localisation sur la surface corticale souffre d´une mauvaise résolution spatiale. Le problème inverse MEG est dit mal-posé et doit de ce fait être régularisé. La parcellisation du cortex en régions de spécificité fonctionnelle proche constitue une régularisation spatiale pertinente du problème inverse MEG. Nous proposons une méthode de parcellisation du cortex entier à partir de la connectivité anatomique cartographiée par imagerie de diffusion. Au sein de chaque aire d´une préparcellisation, la matrice de corrélation entre les profils de connectivité des sources est partitionnée. La parcellisation obtenue est alors mise à jour en testant la similarité des données de diffusion de part et d´autre des frontières de la préparcellisation. C´est à partir de ce résultat que nous contraignons spatialement le problème inverse MEG. Dans ce contexte, deux méthodes sont développées. La première consiste à partitionner l´espace des sources au regard de la parcellisation. L´activité corticale est alors obtenue sur un ensemble de parcelles. Afin de ne pas forcer les sources à avoir exactement la même intensité au sein d´une parcelle, nous développons une méthode alternative introduisant un nouveau terme de régularisation qui, lorsqu´il est minimisé, tend à ce que les sources d´une même parcelle aient des valeurs de reconstruction proches. Nos méthodes de reconstruction sont testées et validées sur des données simulées et réelles. Une application clinique dans le cadre du traitement de données de sujets épileptiques est également réalisée. / Magnetoencephalography (MEG) is a functional non-invasive modality which provides information on the temporal succession of cognitive processes with an excellent time resolution. Unfortunately, spatial resolution is limited due to the ill-posed nature of the MEG inverse problem for estimating source currents from the electromagnetic measurement. Cortex parcellation into regions sharing functional features constitutes a relevant spatial regularization. We propose a whole cortex parcellation method based on the anatomical connectivity mapped by diffusion MRI. Inside areas of a preparcellation, the correlation matrix between connectivity profiles is clustered. The cortex parcellation is then updated testing the similarity of diffusion data on both sides of pre-parcellation boundaries. MEG inverse problem is constrained from this result. Two methods have been developed. The first one is based on the subdivision of source space regarding the parcellation. The cortical activity is obtained on a set of parcels and its analysis is simplified. Not to force sources to have exactly the same value inside a cortical area, we develop an alternative method. We introduce a new regularization term in the MEG inverse problem which constrain sources in a same region to have close values. Our methods are applied on simulated and real subjects. Clinical application is also performed on epileptic data. Each contribution takes part of a pipeline whose each step is detailed to make our works reproducible.

Diffusion magnetic resonance imaging (dMRI) has made a significant breakthrough in neurological disorders and brain research thanks to its exquisite sensitivity to tissue cytoarchitecture. However, as the water diffusion process in neuronal tissues is a complex biophysical phenomena at molecular scale, it is difficult to infer tissue microscopic characteristics on a voxel scale from dMRI data. The major methodological contribution of this thesis is the development of an integrated and generic Monte Carlo simulation framework, 'Diffusion Microscopist Simulator' (DMS), which has the capacity to create 3D biological tissue models of various shapes and properties, as well as to synthesize dMRI data for a large variety of MRI methods, pulse sequence design and parameters. DMS aims at bridging the gap between the elementary diffusion processes occurring at a micrometric scale and the resulting diffusion signal measured at millimetric scale, providing better insights into the features observed in dMRI, as well as offering ground-truth information for optimization and validation of dMRI acquisition protocols for different applications.We have verified the performance and validity of DMS through various benchmark experiments, and applied to address particular research topics in dMRI. Based on DMS, there are two major application contributions in this thesis. First, we use DMS to investigate the impact of finite diffusion gradient pulse duration (delta) on fibre orientation estimation in dMRI. We propose that current practice of using long delta, which is enforced by the hardware limitation of clinical MRI scanners, is actually beneficial for mapping fibre orientations, even though it violates the underlying assumption made in q-space theory. Second, we employ DMS to investigate the feasibility of estimating axon radius using a clinical MRI system. The results suggest that the algorithm for mapping the direct microstructures is applicable to dMRI data acquired from standard MRI scanners.

Bien qu'outil de référence pour le diagnostic et le suivi de la SEP, l'IRM conventionnelle ne reste que modérément corrélée à l'état clinique du patient. Afin de mieux caractériser les altérations pathologiques, nous employons dans ce travail les techniques d'IRM dites non conventionnelles que sont la spectroscopie par résonance magnétique (SRM) et l'IRM de diffusion. Un premier suivi hebdomadaire, a permis de mettre en évidence la sensibilité des métriques de diffusion et la spécificité de la SRM pour détecter les processus initiaux de la formation d'une lésion.Un second suivi a permis de mettre en évidence des modifications de la diffusivité dans plusieurs faisceaux de substance blanche, avec notamment une diminution de la fraction d'anisotropie et une augmentation de diffusivité radiale, s'aggravant avec l'avancée de la maladie et plus marquée dans les formes progressives.Enfin, l'application de la théorie des graphes a permis de caractériser la connectivité cérébrale dans les quatre formes cliniques et d'étudier leur évolution. Cette étude a permis de mettre en évidence des altérations dans tous les phénotypes cliniques, avec notamment une diminution de la densité du réseau cérébral, plus importante dans les formes progressives de la maladie et tendant à s'accentuer avec la progression de la maladie.Ce travail montre la sensibilité des techniques avancées d'IRM pour la caractérisation des altérations pathologiques et de leur évolution dans la SEP / While conventional MRI is the reference tool for the diagnosis and monitoring of MS, it remains only moderately correlated with the patient’s clinical status. In order to better characterize pathological alterations occurring in MS, we use in this work non-conventional MRI techniques, namely magnetic resonance spectroscopy (MRS) and diffusion MRI.A first weekly follow-up revealed the sensitivity of the diffusion metrics and the specificity of the SRM to detect the initial processes of lesion formation.A second follow-up revealed changes in diffusivity in several white matter fiber bundles, including a decrease in fraction of anisotropy and an increase in radial diffusivity, worsening with advancing disease and more marked in the progressive forms.Finally, the application of graph theory allowed to characterize the brain connectivity in the four clinical forms and to study their evolution. This study allowed us to highlight alterations in all the four clinical phenotypes, including a decrease in the cerebral network density, more marked in the progressive forms of the disease and tending to increase with its progression.This work shows the sensitivity of advanced MRI techniques for the characterization of pathological alterations and their evolution in MS

L'imagerie pondérée en diffusion est un type d'acquisition IRM spécifique basé sur la direction de diffusion des molécules d'eau dans le cerveau. Cela permet, au moyen de plusieurs acquisitions, de modéliser la microstructure du cerveau, comme la matière blanche qui à une taille très inférieur à la résolution du voxel. L'obtention d'un grand nombre d'images nécessite, pour un usage clinique, des techniques d'acquisition ultra rapide tel que l'imagerie parallèle. Malheureusement, ces images sont entachées de large distorsions. Nous proposons une méthode de recalage par blocs basée sur l'acquisition d'images avec des directions de phase d'encodage opposées. Cette technique spécialement conçue pour des images écho planaires, mais qui peut être générique, corrige les images de façon robuste tout en fournissant un champs de déformation. Cette transformation est applicable à une série entière d'image de diffusion à partir d'une seule image b 0 renversée, ce qui permet de faire de la correction de distorsion avec un temps d'acquisition supplémentaire minimal. Cet algorithme de recalage, qui a été validé à la fois sur des données synthétiques et cliniques, est disponible avec notre programme de traitement d'images Anima. A partir de ces images de diffusion, nous sommes capable de construire des modèles de diffusion multi-compartiment qui représentent la microstructure complexe du cerveau. Pour pouvoir produire des analyses statistiques sur ces modèles, nous devons être capable de faire du recalage, du moyennage, ou encore de créer un atlas d'images. Nous proposons une méthode générale pour interpoler des modèles multi-compartiment comme un problème de simplification basé sur le partitionnement spectral. Cette technique qui est adaptable pour n'importe quel modèle, a été validé à la fois sur des données synthétiques et réelles. Ensuite à partir d'une base de données recalée, nous faisons des analyses statistiques en extrayant des paramètres au niveau du voxel. Une tractographie, spécifiquement conçue pour les modèles multi-compartiment, est aussi utilisée pour faire des analyses en suivant les fibres de matière blanche. Ces outils sont conçus et appliqués à des données réelles pour contribuer à la recherche de biomarqueurs pour les pathologies cérébrales. / Diffusion weighted imaging (DWI) is a specific type of MRI acquisition based on the direction of diffusion of the brain water molecule. Its allow, through several acquisitions, to model brain microstructure, as white matter, which are significantly smaller than the voxel-resolution. To acquire a large number of images in a clinical use, very-fast acquisition technique are required as single-shot imaging, however these acquisitions suffer local large distortions. We propose a Block-Matching registration method based on a the acquisition of images with opposite phase-encoding directions (PED). This technique specially designs for Echo-Planar Images (EPI), but which could be generic, robustly correct images and provide a deformation field. This field is applicable to an entire DWI series from only one reversed b 0 allowing distortion correction with a minimal time acquisition cost. This registration algorithm has been validated both on a phantom data set and on in-vivo data and is available in our source medical image processing toolbox Anima. From these diffusion images, we are able to construct multi-compartments models (MCM) which could represented complex brain microstructure. We need to do registration, average, create atlas on these MCM to be able to make studies and produce statistic analysis. We propose a general method to interpolate MCM as a simplification problem based on spectral clustering. This technique, which is adaptable for any MCM, has been validated for both synthetic and real data. Then, from a registered dataset, we made analysis at a voxel-level doing statistic on MCM parameters. Specifically design tractography can also be perform to make analysis, following tracks, based on individual compartment. All these tools are designed and used on real data and contribute to the search of biomakers for brain diseases.