As imagens por ressonância magnéticas não indispensáveis no diagnóstico e tratamento de tumores do encéfalo devido ao seu alto grau de detalhamento anatômico. A tarefa de segmenta¸cão da região tumoral, nestas, permite uma análise quantitativa mais precisa, viabilizando um melhor acompanhamento da evolução/regressão da doença. Porém, a realização manual de tal trabalho é cansativa e apresenta diversas desvantagens que a tornam proibitiva, fazendo com que nao haja muitos médicos dispostos a realizá-la rotineiramente. Neste trabalho é proposto um sistema para segmenta¸cão automática de tumores do encéfalo. O sistema emprega parâmetros de textura de naturezas diversas, como estatísticos, baseados em modelo, e baseados em transformada, os quais são extraídos de diferentes tipos de imagem comuns à pratica médica (T1, T1 com contraste e FLAIR). As técnicas de análise de textura são capazes de detectar alterações mínimas nos tecidos, às vezes imperceptíveis à visão humana, fato que motiva sua adoção; e podem ser complementadas por informações adicionais como valores de intensidade. O sistema proposto conta com quatro etapas básicas: pré-processamento, extração de características, segmentação e pós-processamento; e baseia-se no uso de uma máquina de vetor de suporte para classificação dos pixeis. Os resultados obtidos mostram que o sistema apresenta uma taxa média de acerto elevada, comparável aos resultados encontrados em trabalhos relacionados, sendo capaz de localizar e delimitar a região tumoral sem necessidade de interação com o usuário. A quantificação dos resultados foi realizada utilizando-se métricas de artigos encontrados na literatura. / Magnetic resonance images are essential in the diagnosing and treatment of brain tumors due to its high amount of anatomic details. The task of segmenting brain tumor regions in these images makes more exact quantitative analysis feasible, allowing a better tracking of the evolution/regression of the disease. Nevertheless, the execution of such task is burdensome, featuring several drawbacks that turns it into a prohibitive one, and makes many doctors unwilling to put it into practice. In this work an automatic brain tumor segmentation system is proposed, in which several types of texture parameters such as statistical, model based and transform based, are applied. Those parameters are extracted from different, extensively used, types of magnetic resonance images (T1, T1 with contrast and FLAIR). Texture analysis techniques are capable of detecting tiny changes in underlying tissue, which are sometimes imperceptible to the human vision, fact that motivates its adoption here. Texture features can also be completed by other kinds of characteristics, such as pixel intensity. The proposed system comprises four basic steps: pre-processing, feature extraction, segmentation, and post-processing, and is based on a support vector machine for pixel classification. Final results shows that the system archived high success rates, which are comparable to results found in related works, and that it was able to locate and delimit tumor areas without any user interaction. For the quantification of the results, some metrics found in papers presented in the literature were adopted.

A displasia cortical focal (DCF) é uma das causas mais frequentes de epilepsia refratária. Na clínica, diferentes informações são usadas para localizar o foco epileptogênico, mas nenhum método é autossuficiente para evidenciar o local original das crises, associado com a presença da DCF. Embora haja relatos na literatura indicando alterações no padrão de distribuição de tons de cinza e morfologia dos voxels decorrentes da DCF, algumas limitações dos métodos desenvolvidos ainda impedem a utilização clínica. Nossa proposta foi investigar a capacidade de identificar DCF através de análises de espessura cortical e padrões de textura em imagens estruturais de Ressonância Magnética (RM), validando os métodos desenvolvidos a partir uma base de imagens retrospectiva, cujo tecido epileptogênico já havia sido ressecado e a DCF confirmada em análise histológica. A caracterização das DCF foi feita a partir da segmentação automática de tecido cortical saudável em conjunto com a segmentação manual da DCF feita por um especialista, e consiste na geração de mapas de característica e extração de valores de distribuições para comparação em análise estatística. Investigamos também a eficácia da detecção de DCF através do uso de algoritmos de aprendizado de máquina para classificação automática. Obtivemos precisão 0,81 e sensitividade 0,87, colocando o método desenvolvido em par com outros métodos presentes na literatura. Entretanto, foi identificada uma grande dependência do desempenho de métodos de pré-processamento, como corregistro e segmentação automática. / Focal Cortical Dysplasia (FCD) is one of the most frequent causes of refractory epilepsy. In clinical procedures, the information gathered from different techniques is used in order to locate the epileptogenic focus, associated with the presence of FCD. However, there is no self sufficient method to evidence the presence and location of such lesions and especially its extension. Although there are reports indicating change in gray scale intensity patterns and voxel morphology in the presence of DCF, limitations in developed methods still prevent their clinical use. Our proposal was to investigate the capability of identifying FCD through cortical thickness and texture patter analysis in structural MRI images, validating developed methods by utilizing a retrospective base of images from patients that were subjected to surgery, with the FCD being confirmed in histological analysis. Characterization of FCD was achieved from automatic segmentation of healthy cortex and manual segmentation of FCD tissue made by an specialist, and consists in the generation of texture or structural feature maps and comparison of distribution values in healthy or FCD tissue with statistical analysis. We also investigate the efficiency of FCD detection with Machine Learning automatic classification, obtaining precision of 0,81 and sensitivity of 0,87, placing our method on par with other methods in the literature. However, there is a major performance dependency of proposed method with pre-processing steps, like registration and automatic segmentation.

Medical studies increasingly use multi-modality imaging, producing multidimensional data that bring additional information that are also challenging to process and interpret. As an example, for predicting salvageable tissue, ischemic studies in which combinations of different multiple MRI imaging modalities (DWI, PWI) are used produced more conclusive results than studies made using a single modality. However, the multi-modality approach necessitates the use of more advanced algorithms to perform otherwise regular image processing tasks such as filtering, segmentation and clustering. A robust method for addressing the problems associated with processing data obtained from multi-modality imaging is Mean Shift which is based on feature space analysis and on non-parametric kernel density estimation and can be used for multi-dimensional filtering, segmentation and clustering. In this thesis, we sought to optimize the mean shift process by analyzing the factors that influence it and optimizing its parameters. We examine the effect of noise in processing the feature space and how Mean Shift can be tuned for optimal de-noising and also to reduce blurring. The large success of Mean Shift is mainly due to the intuitive tuning of bandwidth parameters which describe the scale at which features are analyzed. Based on univariate Plug-In (PI) bandwidth selectors of kernel density estimation, we propose the bandwidth matrix estimation method based on multi-variate PI for Mean Shift filtering. We study the interest of using diagonal and full bandwidth matrix with experiment on synthesized and natural images. We propose a new and automatic volume-based segmentation framework which combines Mean Shift filtering and Region Growing segmentation as well as Probability Map optimization. The framework is developed using synthesized MRI images as test data and yielded a perfect segmentation with DICE similarity measurement values reaching the highest value of 1. Testing is then extended to real MRI data obtained from animals and patients with the aim of predicting the evolution of the ischemic penumbra several days following the onset of ischemia using only information obtained from the very first scan. The results obtained are an average DICE of 0.8 for the animal MRI image scans and 0.53 for the patients MRI image scans; the reference images for both cases are manually segmented by a team of expert medical staff. In addition, the most relevant combination of parameters for the MRI modalities is determined.

Multi-spectral images have the advantage of providing complementary information to resolve some ambiguities. But, the challenge is how to make use of the multi-spectral images effectively. In this thesis, our study focuses on the fusion of multi-spectral images by extracting the most useful features to obtain the best segmentation with the least cost in time. The Support Vector Machine (SVM) classification integrated with a selection of the features in a kernel space is proposed. The selection criterion is defined by the kernel class separability. Based on this SVM classification, a framework to follow up brain tumor evolution is proposed, which consists of the following steps: to learn the brain tumors and select the features from the first magnetic resonance imaging (MRI) examination of the patients; to automatically segment the tumor in new data using a multi-kernel SVM based classification; to refine the tumor contour by a region growing technique; and to possibly carry out an adaptive training. The proposed system was tested on 13 patients with 24 examinations, including 72 MRI sequences and 1728 images. Compared with the manual traces of the doctors as the ground truth, the average classification accuracy reaches 98.9%. The system utilizes several novel feature selection methods to test the integration of feature selection and SVM classifiers. Also compared with the traditional SVM, Fuzzy C-means, the neural network and an improved level set method, the segmentation results and quantitative data analysis demonstrate the effectiveness of our proposed system.

A displasia cortical focal (DCF) é uma das causas mais frequentes de epilepsia refratária. Na clínica, diferentes informações são usadas para localizar o foco epileptogênico, mas nenhum método é autossuficiente para evidenciar o local original das crises, associado com a presença da DCF. Embora haja relatos na literatura indicando alterações no padrão de distribuição de tons de cinza e morfologia dos voxels decorrentes da DCF, algumas limitações dos métodos desenvolvidos ainda impedem a utilização clínica. Nossa proposta foi investigar a capacidade de identificar DCF através de análises de espessura cortical e padrões de textura em imagens estruturais de Ressonância Magnética (RM), validando os métodos desenvolvidos a partir uma base de imagens retrospectiva, cujo tecido epileptogênico já havia sido ressecado e a DCF confirmada em análise histológica. A caracterização das DCF foi feita a partir da segmentação automática de tecido cortical saudável em conjunto com a segmentação manual da DCF feita por um especialista, e consiste na geração de mapas de característica e extração de valores de distribuições para comparação em análise estatística. Investigamos também a eficácia da detecção de DCF através do uso de algoritmos de aprendizado de máquina para classificação automática. Obtivemos precisão 0,81 e sensitividade 0,87, colocando o método desenvolvido em par com outros métodos presentes na literatura. Entretanto, foi identificada uma grande dependência do desempenho de métodos de pré-processamento, como corregistro e segmentação automática. / Focal Cortical Dysplasia (FCD) is one of the most frequent causes of refractory epilepsy. In clinical procedures, the information gathered from different techniques is used in order to locate the epileptogenic focus, associated with the presence of FCD. However, there is no self sufficient method to evidence the presence and location of such lesions and especially its extension. Although there are reports indicating change in gray scale intensity patterns and voxel morphology in the presence of DCF, limitations in developed methods still prevent their clinical use. Our proposal was to investigate the capability of identifying FCD through cortical thickness and texture patter analysis in structural MRI images, validating developed methods by utilizing a retrospective base of images from patients that were subjected to surgery, with the FCD being confirmed in histological analysis. Characterization of FCD was achieved from automatic segmentation of healthy cortex and manual segmentation of FCD tissue made by an specialist, and consists in the generation of texture or structural feature maps and comparison of distribution values in healthy or FCD tissue with statistical analysis. We also investigate the efficiency of FCD detection with Machine Learning automatic classification, obtaining precision of 0,81 and sensitivity of 0,87, placing our method on par with other methods in the literature. However, there is a major performance dependency of proposed method with pre-processing steps, like registration and automatic segmentation.

OBJETIVO: O objetivo deste estudo é descrever as características do sinal nas lesões císticas da neurocisticercose nas imagens ponderadas em difusão e os metabólitos encontrados na espectroscopia de prótons. MATERIAL E MÉTODOS: Estudaram-se 38 pacientes (39 lesões) com neurocisticercose, usando-se difusão e espectroscopia de prótons. Os exames foram realizados em um magneto de 1,5 T (Signa Horizon LX: GE Medical Systems). A difusão foi realizada no plano axial, com múltiplos cortes com seqüência eco planar. A espectroscopia de prótons utilizou a seqüência PRESS (point-resolved spectroscopy) com TR of 1500 ms e TE de 30/135 ms. RESULTADOS: Os cistos apresentaram intensidade de sinal similar a do líquido cefalorraqueano (LCR) na difusão e valores de CDA sobreponíveis, variando de 1,36 a 3,18 x 10-3 mm2/s. Os picos detectáveis na espectroscopia foram lactato (96,3%), succinato (48%), alanina (40%), lipídeos (15%), aminoácidos citosólicos (7,5%) e acetato (3,7%). CONCLUSÃO: As lesões císticas da neurocisticercose apresentaram hipossinal na difusão e os picos encontrados na espectroscopia de prótons, em ordem decrescente de freqüência, foram lactato, succinato, alanina, lipídeos, aminoácidos citosólicos e acetato / PURPOSE: The objective of this study is to describe the signal behavior of cystic neurocysticercotic lesions on diffusion-weighted imaging (DWI) and single voxel proton spectroscopy findings. MATERIALS AND METHODS: We studied 38 patients (39 lesions) with neurocysticercosis, using diffusion-weighted images and proton MR spectroscopy. The examinations were performed on a 1.5 T scanner (Signa Horizon LX: GE Medical Systems). DWI was performed in the axial plane, using a multisection single shot echo planar pulse sequence. The single voxel proton spectroscopy technique used was the point-resolved spectroscopy (PRESS) sequence with a TR of 1500 ms, short and long TE of 30/135 ms. RESULTS: The cysts presented similar signal intensity to the CSF on DWI, with comparable ADC values, ranging from 1.36 to 3.18 x 10-3 mm2/s. The detectable peaks were lactate (96.3%), succinate (48%), alanine (40%), lipids (15%), cytosolic amino acids (7.5%) and acetate (3.7%). CONCLUSION: The cysts of neurocysticercosis presented hyposignal on DWI and peaks of lactate, succinate, alanine, lipids, cytosolic amino acids and acetate in proton spectroscopy, in decreasing order of frequency

O estudo analisa experimentalmente a integração tendinosa no túnel ósseo femoral na reconstrução do ligamento cruzado anterior do joelho esquerdo com o tendão do músculo semitendíneo autólogo, utilizando imagens de ressonância magnética, teste biomecânico e análise histológica em 15 coelhos da raça Nova Zelândia. Após os períodos de quatro, oito e doze semanas do procedimento cirúrgico, os animais foram submetidos ao exame de ressonância magnética para avaliar o túnel femoral dos joelhos. A seguir os animais foram eutanasiados e os joelhos foram submetidos a testes de tração em uma máquina de ensaios mecânicos Kratos para verificar a integração do enxerto nos túneis e a exame histológico do túnel femoral. A análise dos resultados demonstrou integração mecânica do tendão no túnel femoral a partir da 4ª semana em todos os animais estudados e observou-se na histologia e nas imagens da ressonância magnética alterações do enxerto e da área ao redor de forma heterogênea, sugerindo um processo de cicatrização do tendão-osso, porém não se pode afirmar que ocorria a integração até a 12 semanas. / The purpose was to verify the incorporation (healing) of the graft of the semitendinous tendon into the femoral bone tunnel after an anterior cruciate ligament reconstruction, and verify the post operative evolution of the biomechanical histology and magnetic resonance image analysis of the graft into the femoral bone tunnel. Fifteen New Zealand white rabbits were submitted to an intra-articular anterior cruciate ligament reconstruction in the left knee, using semitendinous tendon autograft. The rabbits were submitted to an magnetic resonance image at 4, 8 and 12 weeks after surgery, after which they were euthanized. The left knee of each rabbit was disarticulated and the anterior cruciate ligament reconstruction was tested for the biomechanical properties and histological analysis of the femoral tunnel. Every rabbit knee showed incorporation of the tendon at the femoral tunnel as of the fourth week in all of the knees studied. After the fourth week signs of integration occurred in the histological analysis and heterogeneous alterations in the magnetic resonance image of the graft and the surrounding areas, which suggests a healing process. Despite the biomechanical incorporation of the graft in the femoral bone tunnel after the fourth week it was not possible to affirm that there occurred incorporation of the graft until the completion of 12 weeks in histological and magnetic resonance image analysis.

Transkranielle Elektrostimulation (tES) beschreibt eine Gruppe von Hirnstimulationstechniken, die einen schwachen elektrischen Strom über zwei nicht-invasiv am Kopf angebrachten Elektroden applizieren. Handelt es sich dabei um einen Gleichstrom, spricht man von transkranieller Gleichstromstimulation, auch tDCS abgekürzt. Die allgemeine Zielstellung aller Hirnstimulationstechniken ist Hirnfunktion durch ein Verstärken oder Dämpfen von Hirnaktivität zu beeinflussen. Unter den Stimulationstechniken wird die transkranielle Gleichstromstimulation als ein adjuvantes Werkzeug zur Unterstützung der mikroskopischen Reorganisation des Gehirnes in Folge von Lernprozessen und besonders der Rehabilitationstherapie nach einem Schlaganfall untersucht. Aktuelle Herausforderungen dieser Forschung sind eine hohe Variabilität im erreichten Stimulationseffekt zwischen den Probanden sowie ein unvollständiges Verständnis des Zusammenspiels der der Stimulation zugrundeliegenden Mechanismen. Als Schlüsselkomponente für das Verständnis der Stimulationsmechanismen wird das zwischen den Elektroden im Kopf des Probanden aufgebaute elektrische Feld erachtet. Einem grundlegenden Konzept folgend wird angenommen, dass Hirnareale, die einer größeren elektrischen Feldstärke ausgesetzt sind, ebenso einen höheren Stimulationseffekt erfahren. Damit kommt der Positionierung der Elektroden eine entscheidende Rolle für die Stimulation zu. Allerdings verteilt sich das elektrische Feld wegen des heterogenen elektrischen Leitfähigkeitsprofil des menschlichen Kopfes nicht uniform im Gehirn der Probanden. Außerdem ist das Verteilungsmuster auf Grund anatomischer Unterschiede zwischen den Probanden verschieden. Die triviale Abschätzung der Ausbreitung des elektrischen Feldes anhand der bloßen Position der Stimulationselektroden ist daher nicht ausreichend genau für eine zielgerichtete Stimulation. Computerbasierte, biophysikalische Simulationen der transkraniellen Elektrostimulation ermöglichen die individuelle Approximation des Verteilungsmusters des elektrischen Feldes in Probanden basierend auf deren medizinischen Bildgebungsdaten. Sie werden daher zunehmend verwendet, um tDCS-Anwendungen zu planen und verifizieren, und stellen ein wesentliches Hilfswerkzeug auf dem Weg zu individualisierter Schlaganfall-Rehabilitationstherapie dar. Softwaresysteme, die den dahinterstehenden individualisierten Verarbeitungsprozess erleichtern und für ein breites Feld an Forschern zugänglich machen, wurden in den vergangenen Jahren für den Anwendungsfall in gesunden Erwachsenen entwickelt. Jedoch bleibt die Simulation von Patienten mit krankhaftem Hirngewebe und strukturzerstörenden Läsionen eine nicht-triviale Aufgabe. Daher befasst sich das hier vorgestellte Projekt mit dem Aufbau und der praktischen Anwendung eines Arbeitsablaufes zur Simulation transkranieller Elektrostimulation. Dabei stand die Anforderung im Vordergrund medizinische Bildgebungsdaten insbesondere neurologischer Patienten mit krankhaft verändertem Hirngewebe verarbeiten zu können. Der grundlegende Arbeitsablauf zur Simulation wurde zunächst für gesunde Erwachsene entworfen und validiert. Dies umfasste die Zusammenstellung medizinischer Bildverarbeitungsalgorithmen zu einer umfangreichen Verarbeitungskette, um elektrisch relevante Strukturen in den Magnetresonanztomographiebildern des Kopfes und des Oberkörpers der Probanden zu identifizieren und zu extrahieren. Die identifizierten Strukturen mussten in Computermodelle überführt werden und das zugrundeliegende, physikalische Problem der elektrischen Volumenleitung in biologischen Geweben mit Hilfe numerischer Simulation gelöst werden. Im Verlauf des normalen Alterns ist das Gehirn strukturellen Veränderungen unterworfen, unter denen ein Verlust des Hirnvolumens sowie die Ausbildung mikroskopischer Veränderungen seiner Nervenfaserstruktur die Bedeutendsten sind. In einem zweiten Schritt wurde der Arbeitsablauf daher erweitert, um diese Phänomene des normalen Alterns zu berücksichtigen. Die vordergründige Herausforderung in diesem Teilprojekt war die biophysikalische Modellierung der veränderten Hirnmikrostruktur, da die resultierenden Veränderungen im Leitfähigkeitsprofil des Gehirns bisher noch nicht in der Literatur quantifiziert wurden. Die Erweiterung des Simulationsablauf zeichnete sich vorrangig dadurch aus, dass mit unsicheren elektrischen Leitfähigkeitswerten gearbeitet werden konnte. Damit war es möglich den Einfluss der ungenau bestimmbaren elektrischen Leitfähigkeit der verschiedenen biologischen Strukturen des menschlichen Kopfes auf das elektrische Feld zu ermitteln. In einer Simulationsstudie, in der Bilddaten von 88 Probanden einflossen, wurde die Auswirkung der veränderten Hirnfaserstruktur auf das elektrische Feld dann systematisch untersucht. Es wurde festgestellt, dass sich diese Gewebsveränderungen hochgradig lokal und im Allgemeinen gering auswirken. Schließlich wurden in einem dritten Schritt Simulationen für Schlaganfallpatienten durchgeführt. Ihre großen, strukturzerstörenden Läsionen wurden dabei mit einem höheren Detailgrad als in bisherigen Arbeiten modelliert und physikalisch abermals mit unsicheren Leitfähigkeiten gearbeitet, was zu unsicheren elektrischen Feldabschätzungen führte. Es wurden individuell berechnete elektrische Felddaten mit der Hirnaktivierung von 18 Patienten in Verbindung gesetzt, unter Berücksichtigung der inhärenten Unsicherheit in der Bestimmung der elektrischen Felder. Das Ziel war zu ergründen, ob die Hirnstimulation einen positiven Einfluss auf die Hirnaktivität der Patienten im Kontext von Rehabilitationstherapie ausüben und so die Neuorganisierung des Gehirns nach einem Schlaganfall unterstützen kann. Während ein schwacher Zusammenhang hergestellt werden konnte, sind weitere Untersuchungen nötig, um diese Frage abschließend zu klären.:Kurzfassung Abstract Contents 1 Overview 2 Anatomical structures in magnetic resonance images 2 Anatomical structures in magnetic resonance images 2.1 Neuroanatomy 2.2 Magnetic resonance imaging 2.3 Segmentation of MR images 2.4 Image morphology 2.5 Summary 3 Magnetic resonance image processing pipeline 3.1 Introduction to human body modeling 3.2 Description of the processing pipeline 3.3 Intermediate and final outcomes in two subjects 3.4 Discussion, limitations & future work 3.5 Conclusion 4 Numerical simulation of transcranial electric stimulation 4.1 Electrostatic foundations 4.2 Discretization of electrostatic quantities 4.3 The numeric solution process 4.4 Spatial discretization by volume meshing 4.5 Summary 5 Simulation workflow 5.1 Overview of tES simulation pipelines 5.2 My implementation of a tES simulation workflow 5.3 Verification & application examples 5.4 Discussion & Conclusion 6 Transcranial direct current stimulation in the aging brain 6.1 Handling age-related brain changes in tES simulations 6.2 Procedure of the simulation study 6.3 Results of the uncertainty analysis 6.4 Findings, limitations and discussion 7 Transcranial direct current stimulation in stroke patients 7.1 Bridging the gap between simulated electric fields and brain activation in stroke patients 7.2 Methodology for relating simulated electric fields to functional MRI data 7.3 Evaluation of the simulation study and correlation analysis 7.4 Discussion & Conclusion 8 Outlooks for simulations of transcranial electric stimulation List of Figures List of Tables Glossary of Neuroscience Terms Glossary of Technical Terms Bibliography / Transcranial electric current stimulation (tES) denotes a group of brain stimulation techniques that apply a weak electric current over two or more non-invasively, head-mounted electrodes. When employing a direct-current, this method is denoted transcranial direct current stimulation (tDCS). The general aim of all tES techniques is the modulation of brain function by an up- or downregulation of brain activity. Among these, transcranial direct current stimulation is investigated as an adjuvant tool to promote processes of the microscopic reorganization of the brain as a consequence of learning and, more specifically, rehabilitation therapy after a stroke. Current challenges of this research are a high variability in the achieved stimulation effects across subjects and an incomplete understanding of the interplay between its underlying mechanisms. A key component to understanding the stimulation mechanism is considered the electric field, which is exerted by the electrodes and distributes in the subjects' heads. A principle concept assumes that brain areas exposed to a higher electric field strength likewise experience a higher stimulation. This attributes the positioning of the electrodes a decisive role for the stimulation. However, the electric field distributes non-uniformly across subjects' brains due to the heterogeneous electrical conductivity profile of the human head. Moreover, the distribution pattern is variable between subjects due to their individual anatomy. A trivial estimation of the distribution of the electric field solely based on the position of the stimulating electrodes is, therefore, not precise enough for a well-targeted stimulation. Computer-based biophysical simulations of transcranial electric stimulation enable the individual approximation of the distribution pattern of the electric field in subjects based on their medical imaging data. They are, thus, increasingly employed for the planning and verification of tDCS applications and constitute an essential tool on the way to individualized stroke rehabilitation therapy. Software pipelines facilitating the underlying individualized processing for a wide range of researchers have been developed for use in healthy adults over the past years, but, to date, the simulation of patients with abnormal brain tissue and structure disrupting lesions remains a non-trivial task. Therefore, the presented project was dedicated to establishing and practically applying a tES simulation workflow. The processing of medical imaging data of neurological patients with abnormal brain tissue was a central requirement in this process. The basic simulation workflow was first designed and validated for the simulation of healthy adults. This comprised compiling medical image processing algorithms into a comprehensive workflow to identify and extract electrically relevant physiological structures of the human head and upper torso from magnetic resonance images. The identified structures had to be converted to computational models. The underlying physical problem of electric volume conduction in biological tissue was solved by means of numeric simulation. Over the course of normal aging, the brain is subjected to structural alterations, among which a loss of brain volume and the development of microscopic alterations of its fiber structure are the most relevant. In a second step, the workflow was, thus, extended to incorporate these phenomena of normal aging. The main challenge in this subproject was the biophysical modeling of the altered brain microstructure as the resulting alterations to the conductivity profile of the brain were so far not quantified in the literature. Therefore, the augmentation of the workflow most notably included the modeling of uncertain electrical properties. With this, the influence of the uncertain electrical conductivity of the biological structures of the human head on the electric field could be assessed. In a simulation study, including imaging data of 88 subjects, the influence of the altered brain fiber structure on the electric field was then systematically investigated. These tissue alterations were found to exhibit a highly localized and generally low impact. Finally, in a third step, tDCS simulations of stroke patients were conducted. Their large, structure-disrupting lesions were modeled in a more detailed manner than in previous stroke simulation studies, and they were physically, again, modeled by uncertain electrical conductivity resulting in uncertain electric field estimates. Individually simulated electric fields were related to the brain activation of 18 patients, considering the inherently uncertain electric field estimations. The goal was to clarify whether the stimulation exerts a positive influence on brain function in the context of rehabilitation therapy supporting brain reorganization following a stroke. While a weak correlation could be established, further investigation will be necessary to answer that research question.:Kurzfassung Abstract Contents 1 Overview 2 Anatomical structures in magnetic resonance images 2 Anatomical structures in magnetic resonance images 2.1 Neuroanatomy 2.2 Magnetic resonance imaging 2.3 Segmentation of MR images 2.4 Image morphology 2.5 Summary 3 Magnetic resonance image processing pipeline 3.1 Introduction to human body modeling 3.2 Description of the processing pipeline 3.3 Intermediate and final outcomes in two subjects 3.4 Discussion, limitations & future work 3.5 Conclusion 4 Numerical simulation of transcranial electric stimulation 4.1 Electrostatic foundations 4.2 Discretization of electrostatic quantities 4.3 The numeric solution process 4.4 Spatial discretization by volume meshing 4.5 Summary 5 Simulation workflow 5.1 Overview of tES simulation pipelines 5.2 My implementation of a tES simulation workflow 5.3 Verification & application examples 5.4 Discussion & Conclusion 6 Transcranial direct current stimulation in the aging brain 6.1 Handling age-related brain changes in tES simulations 6.2 Procedure of the simulation study 6.3 Results of the uncertainty analysis 6.4 Findings, limitations and discussion 7 Transcranial direct current stimulation in stroke patients 7.1 Bridging the gap between simulated electric fields and brain activation in stroke patients 7.2 Methodology for relating simulated electric fields to functional MRI data 7.3 Evaluation of the simulation study and correlation analysis 7.4 Discussion & Conclusion 8 Outlooks for simulations of transcranial electric stimulation List of Figures List of Tables Glossary of Neuroscience Terms Glossary of Technical Terms Bibliography

Les maladies cardiovasculaires sont parmi les principales causes de mortalité à l’échelle mondiale. Un certain nombre de maladies cardiaques peuvent être identifiées et localisées par l’analyse du mouvement et de la déformation cardiaques à partir de l’imagerie médicale. Cependant, l’utilisation de ces techniques en routine clinique est freinée par le manque d’outils de quantification efficaces et fiables. Dans cette thèse, nous introduisons un algorithme de quantification appliqué aux images IRM marquées. Nous présentons ensuite un pipeline de simulation qui génère des séquences cardiaques synthétiques (US et IRM). Les principales contributions sont décrites ci-dessous. Tout d’abord, nous avons proposé une nouvelle extension 3D de la méthode de la phase harmonique. Le suivi de flux optique en utilisant la phase a été combiné avec un modèle de régularisation anatomique afin d’estimer les mouvements cardiaques à partir des images IRM marquées. En particulier, des efforts ont été faits pour assurer une estimation précise de la déformation radiale en imposant l’incompressibilité du myocarde. L’algorithme (dénommé HarpAR) a ensuite été évalué sur des volontaires sains et des patients ayant différents niveaux d’ischémie. HarpAR a obtenu la précision de suivi comparable à quatre autres algorithmes de l’état de l’art. Sur les données cliniques, la dispersion des déformations est corrélée avec le degré de fibroses. De plus, les segments ischémiques sont distingués des segments sains en analysant les courbes de déformation. Deuxièmement, nous avons proposé un nouveau pipeline de simulation pour générer des séquences synthétiques US et IRM pour le même patient virtuel. Les séquences réelles, un modèle électromécanique (E/M) et les simulateurs physiques sont combinés dans un cadre unifié pour générer des images synthétiques. Au total, nous avons simulé 18 patients virtuels, chacun avec des séquences synthétiques IRM cine, IRM marquée et US 3D. Les images synthétiques ont été évaluées qualitativement et quantitativement. Elles ont des textures d’images réalistes qui sont similaires aux acquisitions réelles. De plus, nous avons également évalué les propriétés mécaniques des simulations. Les valeurs de la fraction d’éjection et de la déformation locale sont cohérentes avec les valeurs de référence publiées dans la littérature. Enfin, nous avons montré une étude préliminaire de benchmarking en utilisant les images synthétiques. L'algorithme générique gHarpAR a été comparé avec un autre algorithme générique SparseDemons en termes de précision sur le mouvement et la déformation. Les résultats montrent que SparseDemons surclasse gHarpAR en IRM cine et US. En IRM marquée, les deux méthodes ont obtenu des précisions similaires sur le mouvement et deux composants de déformations (circonférentielle et longitudinale). Toutefois, gHarpAR estime la déformation radiale de manière plus précise, grâce à la contrainte d’incompressibilité du myocarde. / Cardiovascular disease is one of the major causes of death worldwide. A number of heart diseases can be diagnosed through the analysis of cardiac images after quantifying shape and function. However, the application of these deformation quantification algorithms in clinical routine is somewhat held back by the lack of a solid validation. In this thesis, we mainly introduce a fast 3D tagged MR quantification algorithm, as well as a novel pipeline for generating synthetic cardiac US and MR image sequences for validation purposes. The main contributions are described below. First, we proposed a novel 3D extension of the well-known harmonic phase tracking method. The point-wise phase-based optical flow tracking was combined with an anatomical regularization model in order to estimate anatomically coherent myocardial motions. In particular, special efforts were made to ensure a reasonable radial strain estimation by enforcing myocardial incompressibility through the divergence theorem. The proposed HarpAR algorithm was evaluated on both healthy volunteers and patients having different levels of ischemia. On volunteer data, the tracking accuracy was found to be as accurate as the best candidates of a recent benchmark. On patient data, strain dispersion was shown to correlate with the extent of transmural fibrosis. Besides, the ischemic segments were distinguished from healthy ones from the strain curves. Second, we proposed a simulation pipeline for generating realistic synthetic cardiac US, cine and tagged MR sequences from the same virtual subject. Template sequences, a state-of-the-art electro-mechanical (E/M) model and physical simulators were combined in a unified framework for generating image data. In total, we simulated 18 virtual patients (3 healthy, 3 dyssynchrony and 12 ischemia), each with synthetic sequences of 3D cine MR, US and tagged MR. The synthetic images were assessed both qualitatively and quantitatively. They showed realistic image textures similar to real acquisitions. Besides, both the ejection fraction and regional strain values are in agreement with reference values published in the literature. Finally, we showed a preliminary benchmarking study using the synthetic database. We performed a comparison between gHarpAR and another tracking algorithm SparseDemons using the virtual patients. The results showed that SparseDemons outperformed gHarpAR in processing cine MR and US images. Regarding tagged MR, both methods obtained similar accuracies on motion and two strain components (circumferential and longitudinal). However, gHarpAR quantified radial strains more accurately, thanks to the myocardial incompressibility constraint. We conclude that motion quantification solutions can be improved by designing them according to the image characteristics of the modality and that a solid evaluation framework can be a key asset in comparing different algorithmic options.

Dans le cadre de cette thèse, nous nous somme focalisés sur deux objectifs complémentaires. Le premier concerne l’évolution de la méthode du Gabarit Déformable Elastique (GDE) pour l’extraction semi-automatique de l’anatomie et du mouvement cardiaque, développée au laboratoire Creatis. Un travail a d’abord été réalisé sur une base de données de 45 patients afin de mettre en évidence les points forts et les points faibles de l’algorithme, notamment la difficulté à suivre des déformations trop importantes ou des formes inhabituelles. Puis, différents types de contraintes ont été intégrées au modèle GDE afin d’en améliorer les performances : prescription locale ou dense de déplacements, directionnalité de la déformation contrainte par celle des fibres. Les contraintes proposées sont évaluées sur des données de synthèse et des données réelles en IRM ciné et de marquage tissulaire acquises chez l’homme. Parallèlement, une étude a été réalisée pour mettre en place la méthodologie nécessaire à l’extraction et l’analyse statistique de la déformation des fibres myocardiques. Ce travail a été effectué en collaboration avec une équipe du Auckland Bioengineering Institute en Nouvelle-Zélande. Un modèle biomécanique par éléments finis intègre la direction principale des fibres en tout point du ventricule gauche issue d’acquisitions en IRM du tenseur de diffusion (IRM-TD) sur coeurs humains ex vivo et le mouvement issu de séquences IRM marquées. Cette combinaison permet l’estimation de la déformation des fibres et sa variation durant le cycle cardiaque. La variabilité dans la déformation des fibres est étudiée statistiquement à travers le croisement d’une base de données IRM-TD et d’une base de données IRM marquées. / In this thesis, we are interested in two complementary goals. First, we have improved the Dynamic Deformable Elastic Template (DET) model, developed at Creatis, for the semi-automatic extraction of the anatomy and cardiac motion. The performance of the method was assessed on a database consisting in 45 patients and yielded fairly accurate results. However, it experienced difficulties when dealing with very large thickening throughout the cardiac cycle. Thus, different type of constraints were integrated to the DET model in order to improve robustness and accuracy : local or dense prescribed displacements, deformations directionally constrained by the fibres. These constraints are evaluated on simulated and real human data, in both cine and tagged MR images. A methodology has also been developed in order to extract and statistically analyse myocardial fibre strain. This work was done in collaboration with a team at the Auckland Bioengineering Institute in New Zealand. A finite elements biomechanical model integrates the principle direction of fibres in the left ventricle from Diffusion Tensor MRI acquisitions on ex vivo human hearts and motion from tagged MRI sequences. Fibre strain and its variation throughout the cardiac cycle were estimated. Variability in fibre strain is statistically studied by joining DT-MRI and tagged MRI databases.