Segmentation of individual facial soft tissues has received relatively little attention in the literature due to the complicated structures of these tissues. There is a need to incorporate the prior information, which is usually in the form of atlases, in the segmentation process. In this thesis we performed several segmentation methods that take advantage of prior knowledge for facial soft tissue segmentation. An atlas based method and three expectation maximization &ndash / Markov random field (EM-MRF) based methods are tested for two dimensional (2D) segmentation of masseter muscle in the face. Atlas based method uses the manually labeled atlases as prior information. We implemented EM-MRF based method in different manners / without prior information, with prior information for initialization and with using labeled atlas as prior information. The differences between these methods and the influence of the prior information are discussed by comparing the results. Finally a new method based on EM-MRF is proposed in this study. In this method we aim to use prior information without performing manual segmentation, which is a very complicated and time consuming task. 10 MRI sets are used as experimental data in this study and leave-one-out technique is used to perform segmentation for all sets. The test data is modeled as a Markov Random Field where unlabeled training data, i.e., other 9 sets, are used as prior information. The model parameters are estimated by the Maximum Likelihood approach when the Expectation Maximization iterations are used to handle hidden labels. The performance of all segmentation methods are computed and compared to the manual segmented ground truth. Then we used the new 2D segmentation method for three dimensional (3D) segmentation of two masseter and two temporalis tissues in each data set and visualize the segmented tissue volumes.

Cette thèse avait deux buts principaux : a) l'implémentation et l'amélioration d'une méthode de calcul de densité volumique du sein (VBD), et b) la proposition d'une mesure d'irradiation utilisable pour l'évaluation du risque individuel en mammographie avec une méthode pour l'estimer. La densité du sein est connue comme indicateur de risque du cancer. Une méthode de quantification objective de la VBD a été développée, à partir d'approches existantes, et améliorée. La méthode a été implémentée pour deux systèmes de mammographie. Elle repose sur l'étalonnage du système de mammographie et la chaîne d'acquisition avec des fantômes équivalents aux tissus mammaires. Une carte de densité est calculée.La contribution majeure de la thèse consiste en une nouvelle méthode de validation, applicable à tout calcul de VBD d'image de mammographie. Elle consiste à comparer les résultats aux valeurs de densité obtenues par des scanners thoraciques pour la même patiente. Cette validation a été appliquée à notre méthode de calcul et nous avons trouvé 10% d'écart moyen entre les deux méthodes, ce qui est comparable aux résultats de l'état de l'art. Pour le risque d'irradiation individuel, nous proposons de remplacer la dose glandulaire moyenne par l'énergie déposée, qui dépend de la quantité et de la distribution du tissu glandulaire, qui est le tissu à risque. L'énergie volumique déposée est calculée par simulation de Monte Carlo. Le VBD, calculé pour l'image de projection à 0° en tomosynthèse, aide à localiser le tissu glandulaire et à attribuer l'énergie déposée dans les tissus différents. Une proposition a été faite pour des fantômes géométriques, un fantôme texturé et un cas de patiente / In this research project the main goals were a) to implement a method for the computation of the volumetric breast density (VBD), and b) to propose an improved quantity for the assessment of individual radiation-induced risk, in particular during mammography, together with a method to quantify it. The breast density is known as a breast cancer risk factor. The objective quantification of the volumetric breast density was developed, based on already published methods, and improved. The method was implemented for two mammography systems. It is based on the calibration of the mammography system acquisition chain with breast equivalent phantoms and computes a breast density map. Our most important contribution resides in a new validation method applicable to any VBD computation, consisting in comparing its results with the VBD obtained from a thorax CT examination for the same patient. This validation method was applied to our VBD computation. We found an average deviation between mammography and CT of less than 10%. Our results are comparable to the state-of-the-art results for other validation methods. For the individual radiation risk, we proposed to replace the average glandular dose by the imparted energy, which depends on the quantity and distribution of the glandular tissue, which is the tissue at risk. The volumetric imparted energy is computed from Monte Carlo simulations. The VBD, computed for the 0° projection of tomosynthesis exams, helps us to localize the glandular tissue and to attribute the imparted energy to the different tissues. A proposition was implemented for geometric phantoms, a textured phantom and a patient case.

Magnetic Resonance Imaging (MRI) is a cornerstone in neuroimaging for studying brain anatomy and functions. Anatomical MRI images, such as T1-weighted (T1-w) scans, allow the non-invasive visualization of the brain tissues, enabling the investigation of the brain morphology and facilitating the diagnosis of both acquired (e.g., tumors, stroke lesions, infections) and congenital (e.g., malformations) brain disorders. T1-w images provide a detailed representation of brain anatomy and accurate differentiation between the main brain structures, such as white matter (WM) and gray matter (GM), therefor they are frequently used in combination with advanced sequences such as diffusion MRI (dMRI) for the computation of the structural connectivity of the brain. In particular, from the processing of dMRI data, it is possible to investigate the structures of WM through tractography techniques, obtaining a virtual representation of the WM pathways called tractogram. Since the tractogram is a collection of digital fibers representing the neuronal axons connecting the brain's cortical areas, it is the fundamental element for studying the brain's structural connectivity. A critical step for processing the tractography data is the accurate labeling of the brain tissues, usually performed through brain tissue segmentation of T1-w images. Even though the gold standard is manual segmentation, it is time-consuming and prone to intra/inter-operator variability. Automated model-based methods produce more consistent and reliable results, however, they struggle with accuracy in the case of pathological brains due to reliance on priors based on normal anatomy. Recently, deep learning (DL) has shown the potential of supervised data-driven approaches for brain tissue segmentation by leveraging the information encoded in the signal intensity of T1-w images. As a first contribution of this thesis, we reported empirical evidence that a data-driven approach is effective for brain tissue segmentation in pathological brains. By implementing a DL network trained on a large dataset of only healthy subjects, we demonstrated improvements in segmenting the brain tissues compared to models based on healthy anatomical priors, especially on severely distorted brains. Additionally, we published a benchmark for enabling an open investigation into improving tissue segmentation of distorted brains, providing a training dataset of about one thousand healthy individuals with T1-w MR images and corresponding brain tissue labels, and a test dataset includes several tens of individuals with severe brain distortions. Another crucial aspect of processing tractography data for brain connectivity analysis is the correct alignment of the WM structures across different subjects or their normalization into a common reference space, usually performed as tractography alignment. The best practice is to perform the registration using T1-w images and then apply the resulting transformation to align the tractography, despite T1-w images lacking fiber orientation information. In light of this, various methods have been proposed to leverage the information of the WM from dMRI data, ranging from scalar diffusion maps to more complex models encoding fiber orientation in the voxels. As a second contribution to the thesis, we provide a comprehensive survey of methods for conducting tractogram alignment. Additionally, we include an empirical study with the results of a quantitative comparison among the main methods for which an implementation is available. From our findings, the use of increasingly complex diffusion models does not significantly improve the alignment of tractograms. Conversely, correspondence methods that use the fibers directly to compute the alignment outperform voxel-based methods, albeit with some limitations: not producing a deformation field, operating in an unsupervised manner, and avoiding using anatomical information. Recently, geometric deep learning (GDL) models have shown promising results in handling non-grid data like tractograms, offering new possibilities for WM structure alignment. The third main contribution of this thesis is implementing a GDL model for tractogram alignment through a supervised approach guided by fiber correspondence. The alignment is predicted as the displacement of fiber points, based on a GDL registration framework that combines graph convolutional networks and differentiable loopy belief propagation, incorporating the definition of fiber structure into the encoding of the graph. Our empirical analysis demonstrates the advantages of utilizing the proposed GDL framework over traditional volumetric registration, showcasing high alignment accuracy, low inference time, and good generalization capabilities. Overall, this thesis advances the methodology for processing MRI data for brain structural connectivity, addressing the challenges of tissue segmentation and tractography alignment, proving the potential of DL approaches also in the case of pathological brains.