In order to analyse body composition of MRI (Magnetic Resonance Imaging) volumes, atlas-based segmentation is often used to retrieve information from specific organs or anatomical regions. The method behind this technique is to use an already segmented image volume, an atlas, to segment a target image volume by registering the volumes to each other. During this registration a deformation field will be calculated, which is applied to a segmented part of the atlas, resulting in the same anatomical segmentation in the target. The drawback with this method is that the quality of the segmentation is highly dependent on the similarity between the target and the atlas, which means that many atlases are needed to obtain good segmentation results in large sets of MRI volumes. One potential solution to overcome this problem is to create the deformation field between a target and an atlas as a sequence of small deformations between more similar bodies.  In this master thesis a new method for atlas-based segmentation has been developed, with the anticipation of obtaining good segmentation results regardless of the level of similarity between the target and the atlas. In order to do so, 4000 MRI volumes were used to create a manifold of human bodies, which represented a large variety of different body types. These MRI volumes were compared to each other and the calculated similarities were saved in matrices called similarity models. Three different similarity measures were used to create the models which resulted in three different versions of the model. In order to test the hypothesis of achieving good segmentation results when the deformation field was constructed as a sequence of small deformations, the similarity models were used to find the shortest path (the path with the least dissimilarity) between a target and an atlas in the manifold.  In order to evaluate the constructed similarity models, three MRI volumes were chosen as atlases and 100 MRI volumes were randomly picked to be used as targets. The shortest paths between these volumes were used to create the deformation fields as a sequence of small deformations. The created fields were then used to segment the anatomical regions ASAT (abdominal subcutaneous adipose tissue), LPT (left posterior thigh) and VAT (visceral adipose tissue). The segmentation performance was measured with Dice Index, where segmentations constructed at AMRA Medical AB were used as ground truth. In order to put the results in relation to another segmentation method, direct deformation fields between the targets and the atlases were also created and the segmentation results were compared to the ground truth with the Dice Index. Two different types of transformation methods, one non-parametric and one affine transformation, were used to create the deformation fields in this master thesis. The evaluation showed that good segmentation results can be achieved for the segmentation of VAT for one of the constructed similarity models. These results were obtained when a non-parametric registration method was used to create the deformation fields. In order to achieve similar results for an affine registration and to improve the segmentation of other anatomical regions, further investigations are needed.

The study of brain nuclei in neuroimaging poses challenges owing to its small size. Many neuroimaging studies have been reported for effectively locating these nuclei and characterizing their functional connectivity with other regions of the brain. Hypothalamus, Locus Coeruleus, and Ventral Tegmental area are such nuclei found in the human brain, which are challenging to visualize owing to their size and lack of tissue contrast with surrounding regions. Resting-state functional magnetic resonance imaging (rsfMRI) analysis on these nuclei enabled researchers to characterize their connectivity with other regions of the brain. An automated method to successfully isolate voxels belonging to these nuclei is still a great challenge in the field of neuroimaging. Atlas-based segmentation is the most common method used to study the anatomy and the functional connectivity of these brain nuclei. However, atlas-based segmentation has shown inconsistency due to variation in brain atlases owing to different population studies. Therefore, in this study, we try to address the research problem of brain nuclei imaging using a clustering-based approach. Clustering-based methods separate of voxels utilizing their structural and functional homogeneity to each other. This type of method can help locate and cluster the voxels belonging to the nuclei. Elimination of erroneous voxels by the use of clustering methods would significantly improve the structural and functional analysis of the nuclei in the human brain. Since several clustering methods are available in neuroimaging studies, the goal of this study is to find a robust model that has less variability across different subjects. Non-parametrical statistical analysis was performed as functional magnetic resonance imaging (fMRI) based studies are corrupted with noise and artefact. Statistical investigation on the fMRI data helps to assess the significant experimental effects.

Quantifying tissue volumes in pediatric brains from magnetic resonance (MR) images can provide insight into etiology and onset of neurological disease. Unbiased volumetric analysis can be applied to large population studies when automated image processing is possible. Standard segmentation strategies using adult atlases fail to account for varying tissue contrasts and types associated with the rapid growth and maturational changes seen in early neurodevelopment. The goal of this project was to develop an automated pipeline and two age-specific atlases capable of providing accurate tissue classification despite these challenges. The automated pipeline consisted of a stepwise initial atlas-to-subject registration, expectation maximization (EM) atlas based segmentation, and a post-processing level set segmentation for improved white/gray matter separation. This level set segmentation is a 3D and multiphase adaptation of a 2D method intended for use on images with the types of intensity Inhomogeneities found in MR images. The initial tissue maps required to determine spatial priors for the one-year-old atlas were created by manually cleaning the results of an adult atlas and the automated pipeline. Additional tissue maps were incrementally added until the spatial priors were sufficiently representative. The neonate atlas was similarly created, starting with the one-year-old atlas.

Deep brain stimulation (DBS) surgery, a type of microelectrode-guided surgery, is an effective treatment for the movement disorders patients that can no longer be treated by medications. New rigid and non-rigid image registration methods were developed for the movement disorders patients that underwent DBS surgery. These new methods help study and analyze the brain shift during the DBS surgery and perform atlas-based segmentation of the deep brain structures for the DBS surgery planning and navigation. A diploë based rigid registration method for the intra-operative brain shift analysis during the DBS surgery was developed. The proposed method for the brain shift analysis ensures rigid registration based on diploë only, which can be treated as a rigid structure as opposed to the brain tissues. The results show that the brain shift during the DBS surgery is comparable to the size of the DBS targets and should not be neglected. This brain shift may further lengthen and complicate the DBS surgery contrary to the common belief that brain shift during the DBS surgery is not considerable. We also developed an integrated electrophysiological and anatomical atlas with eleven deep brain structures segmented by an expert, and electrophysiological data of four implant locations obtained from post-op MRI data of twenty patients that underwent DBS surgery. This atlas MR image is then non-rigidly registered with the pre-operative patient MR image, which provides initial DBS target location along with the segmented deep brain structures that can be used for guidance during the microelectrode mapping of the stereotactic procedure. The atlas based approach predicts the target automatically as opposed to the manual selection currently used. The results showed that 85% of the times, this automatic selection of the target location was closer to the target when compared to currently used technique.

The aim of this thesis is to investigate the shape change in hippocampus due to the atrophy in Alzheimer's disease (AD). To this end, specific algorithms and methodologies were developed to segment the hippocampus from structural magnetic resonance (MR) images and model variations in its shape. We use a multi-atlas based segmentation propagation approach for the segmentation of hippocampus which has been shown to obtain accurate parcellation of brain structures. We developed a supervised method to build a population specific atlas database, by propagating the parcellations from a smaller generic atlas database. Well segmented images are inspected and added to the set of atlases, such that the segmentation capability of the atlas set may be enhanced. The population specific atlases are evaluated in terms of the agreement among the propagated labels when segmenting new cases. Compared with using generic atlases, the population specific atlases obtain a higher agreement when dealing with images from the target population. Atlas selection is used to improve segmentation accuracy. In addition to the conventional selection by image similarity ranking, atlas selection based on maximum marginal relevance (MMR) re-ranking and least angle regression (LAR) sequence are developed for atlas selection. By taking the redundancy among atlases into consideration, diversity criteria are shown to be more efficient in atlas selection which is applicable in the situation where the number of atlases to be fused is limited by the computational resources. Given the segmented hippocampal volumes, statistical shape models (SSMs) of hippocampi are built on the samples to model the shape variation among the population. The correspondence across the training samples of hippocampi is established by a groupwise optimization of the parameterized shape surfaces. The spherical parameterization of the hippocampal surfaces are flatten to facilitate the reparameterization and interpolation. The reparameterization is regularized by viscous fluid, which is solved by a fast implementation based on discrete sine transform. In order to use the hippocampal SSM to describe the shape of an unseen hippocampal surface, we developed a shape parameter estimator based on the expectationmaximization iterative closest points (EM-ICP) algorithm. A symmetric data term is included to achieve the inverse consistency of the transformation between the model and the shape, which gives more accurate reconstruction of the shape from the model. The shape prior modeled by the SSM is used in the maximum a posteriori estimation of the shape parameters, which is shown to enforce the smoothness and avoid the effect of over-fitting. In the study of the hippocampus in AD, we use the SSM to model the hippocampal shape change between the healthy control subjects and patients diagnosed with AD. We identify the regions affected by the atrophy in AD by assessing the spatial difference between the control and AD groups at each corresponding landmark. Localized shape analysis is performed on the regions exhibiting significant inter-group difference, which is shown to improve the discrimination ability of the principal component analysis (PCA) based SSM. The principal components describing the localized shape variability among the population are also shown to display stronger correlation with the decline of episodic memory scores linked to the pathology of hippocampus in AD.

L’objectif de cette thèse est l’étude des changements de la forme de l’hippocampe due à l’atrophie causée par la maladie d’Alzheimer. Pour ce faire, des algorithmes et des méthodes ont été développés pour segmenter l’hippocampe à partir d’imagerie structurelle par résonance magnétique (IRM) et pour modéliser les variations dans sa forme. Nous avons utilisé une méthode de segmentation par propagation de multiple atlas pour la segmentation de l’hippocampe, méthode qui a été démontrée comme étant robuste dans la segmentation des structures cérébrales. Nous avons développé une méthode supervisée pour construire une base de données d’atlas spécifique à la population d’intérêt en propageant les parcellations d’une base de données génériques d’atlas. Les images correctement segmentées sont inspectées et ajoutées à la base de données d’atlas, de manière à améliorer sa capacité à segmenter de nouvelles images. Ces atlas sont évalués en termes de leur accord lors de la segmentation de nouvelles images. Comparé aux atlas génériques, les atlas spécifiques à la population d’intérêt obtiennent une plus grande concordance lors de la segmentation des des images provenant de cette population. La sélection d’atlas est utilisée pour améliorer la précision de la segmentation. La méthode classique de sélection basée sur la similarité des images est ici étendue pour prendre en compte la pertinence marginale maximale (MMR) et la régression des moindres angles (LAR). En prenant en considération la redondance parmi les atlas, des critères de diversité se montrent être plus efficace dans la sélection des atlas dans le cas où seul un nombre limité d’atlas peut-être fusionné. A partir des hippocampes segmentés, des modèles statistiques de la forme (SSM) sont construits afin de modéliser les variations de la forme de l’hippocampe dans la population. La correspondance entre les hippocampes est établie par une optimisation d’ensemble des surfaces paramétriques. Les paramétrages sphériques des surfaces sont aplatis pour faciliter la reparamétrisation et l’interpolation. Le reparamétrage est régularisé par une contrainte de type fluide visqueux, qui est effectué à l’aide d’une implémentation basée sur la transformées en sinus discrète. Afin d’utiliser le SSM pour décrire la forme d’une nouvelle surface hippocampique, nous avons développé un estimateur des paramètres du model de la forme basée sur l’espérance-maximisation de l’algorithme du plus proche voisin itéré (EM-ICP). Un terme de symétrie est inclus pour forcer une consistance entre la transformée directe et inverse entre le modèle et la forme, ce qui permet une reconstruction plus précise de la forme à partir du modèle. La connaissance a priori sur la forme modélisé par le SSM est utilisée dans l’estimation du maximum a posteriori des paramètres de forme. Cette méthode permet de forcer la continuité spatiale et éviter l’effet de sur-apprentissage. Dans l’étude de l’hippocampe dans la maladie d’Alzheimer, nous utilisons le SSM pour modéliser le changement de forme de l’hippocampe entre les sujets sains et des patients souffrant d’Alzheimer. Nous identifions les régions touchées par l’atrophie dans la maladie d’Alzheimer en évaluant la différence entre les groupes de contrôle et ceux d’Alzheimer sur chaque point correspondant sur la surface. L’analyse des changements de la forme est restreinte aux régions présentant des différences significatives entre les groupes, ce qui a pour effet d’améliorer la discrimination basée sur l’analyse en composantes principales (ACP) du SSM. Les composantes principales décrivant la variabilité de la forme à l’intérieur des régions discriminantes ont une corrélation plus fortes avec le déclin des scores de mémoire épisodique liée à la pathologie de l’hippocampe dans la maladie d’Alzheimer. / The aim of this thesis is to investigate the shape change in hippocampus due to the atrophy in Alzheimer’s disease (AD). To this end, specific algorithms and methodologies were developed to segment the hippocampus from structural magnetic resonance (MR) images and model variations in its shape. We use a multi-atlas based segmentation propagation approach for the segmentation of hippocampus which has been shown to obtain accurate parcellation of brain structures. We developed a supervised method to build a population specific atlas database, by propagating the parcellations from a smaller generic atlas database. Well segmented images are inspected and added to the set of atlases, such that the segmentation capability of the atlas set may be enhanced. The population specific atlases are evaluated in terms of the agreement among the propagated labels when segmenting new cases. Compared with using generic atlases, the population specific atlases obtain a higher agreement when dealing with images from the target population. Atlas selection is used to improve segmentation accuracy. In addition to the conventional selection by image similarity ranking, atlas selection based on maximum marginal relevance (MMR) re-ranking and least angle regression (LAR) sequence are developed for atlas selection. By taking the redundancy among atlases into consideration, diversity criteria are shown to be more efficient in atlas selection which is applicable in the situation where the number of atlases to be fused is limited by the computational resources. Given the segmented hippocampal volumes, statistical shape models (SSMs) of hippocampi are built on the samples to model the shape variation among the population. The correspondence across the training samples of hippocampi is established by a groupwise optimization of the parameterized shape surfaces. The spherical parameterization of the hippocampal surfaces are flatten to facilitate the reparameterization and interpolation. The reparameterization is regularized by viscous fluid, which is solved by a fast implementation based on discrete sine transform. In order to use the hippocampal SSM to describe the shape of an unseen hippocampal surface, we developed a shape parameter estimator based on the expectationmaximization iterative closest points (EM-ICP) algorithm. A symmetric data term is included to achieve the inverse consistency of the transformation between the model and the shape, which gives more accurate reconstruction of the shape from the model. The shape prior modeled by the SSM is used in the maximum a posteriori estimation of the shape parameters, which is shown to enforce the smoothness and avoid the effect of over-fitting. In the study of the hippocampus in AD, we use the SSM to model the hippocampal shape change between the healthy control subjects and patients diagnosed with AD. We identify the regions affected by the atrophy in AD by assessing the spatial difference between the control and AD groups at each corresponding landmark. Localized shape analysis is performed on the regions exhibiting significant inter-group difference, which is shown to improve the discrimination ability of the principal component analysis (PCA) based SSM. The principal components describing the localized shape variability among the population are also shown to display stronger correlation with the decline of episodic memory scores linked to the pathology of hippocampus in AD.