Après une déprivation sensorielle ou après une réhabilitation sensorielle périphérique, les aires sensori-motrices primaires du cerveau se réorganisent anatomiquement et fonctionnellement. Dans le cas du transfert tendineux, une réhabilitation des fonctions manuelles est opérée grâce à une inversion de certaines fonctions de flexion en extension. Chez les patients opérés, la commande centrale de ces mouvements élémentaires repose sur des processus de plasticité cérébrale, processus encore mal connus, lents qui se stabilisent en une année et qui présentent une co-occurrence avec la récupération manuelle. Nous faisons l'hypothèse que l'IRM fonctionnelle de haute résolution permet de détecter la plasticité cérébrale liée à cette récupération motrice. Cette thèse a pour but de mettre en place des méthodes de suivi longitudinal individuel à partir des images fonctionnelles cérébrales des patients pour étudier la plasticité cérébrale lors de la récupération fonctionnelle. On a d’abord cherché à obtenir des cartes de référence de la motricité fine de la main sur un groupe de sujets contrôles. Ces cartes sont ensuite utilisées pour évaluer, à partir des données obtenues lors du suivi longitudinal de patients, la plasticité cérébrale liée au transfert tendineux. Pour mieux comprendre la représentation des mouvements de flexion et d'extension de la main dans le cortex moteur primaire, nous avons mis en place une étude en IRM fonctionnelle de haute résolution à 3T sur un groupe de 13 sujets sains. Pour généraliser les résultats individuels obtenus pour chaque mouvement au niveau de la population, il faut pouvoir réaligner les cerveaux individuels dans un référentiel commun afin d'effectuer une analyse statistique sur le groupe. Or, la variabilité anatomique inter-individuelle des aires motrices primaires constituait un verrou scientifique car celles-ci ne se réalignaient pas correctement par les méthodes classiques. Dans un premier temps, nous avons donc évalué différentes méthodes de recalage non-linéaires au niveau du cortex moteur pour assurer un alignement inter-individuel robuste. Nous avons comparé une méthode locale, Demons qui permet des déformations locales très importantes des images, une méthode classique non-linéaire (SPM), qui permet un alignement global des images, et une méthode de recalage global difféomorphique (DARTEL) utilisant de nombreux degrés de liberté. De plus, nous avons appliqué une méthode récente basée sur l’alignement des sillons corticaux au préalable à l’utilisation de DARTEL (DISCO+DARTEL) en partenariat avec l’équipe du LSIS, (CNRS Marseille) qui ajoute un caractère local à la technique globale de recalage. Nous montrons que cette dernière méthode produit le meilleur alignement sur des critères à la fois anatomiques (distance de Hausdorff, superposition des masques de tissus, et caractéristiques du champ de déformations) et fonctionnels (précision de la localisation et robustesse statistique des activations détectées). Nous avons alors pu établir des cartes de référence en haute résolution des mouvements élémentaires de la main, selon la dominance manuelle, selon la direction du mouvement et selon le segment de la main. Ces cartes présentent de multiples foyers d’activation et un chevauchement important dans la région dite du « hand-knob ». Dans un deuxième temps, nous avons évalué la robustesse de nos résultats, en réalisant une deuxième série d’examens sur les mêmes sujets contrôles (test-retest). Deux patients ont été suivis longitudinalement: avant la chirurgie (M0), à 1 mois (M1), à 3 (M3), 6 (M6) et 12 mois (M12) après la chirurgie. On a observé le recrutement au cours de la récupération de la SMA, de M1 et S1 ipsilatéral, de S1 contralatéral et des lobules V et VI du cervelet. Par rapport aux sujets sains, les patients mettent en place une réorganisation compensatoire au cours de la récupération fonctionnelle. / After a sensory deprivation or a peripheral sensory rehabilitation, primary sensorimotor areas undergo anatomical and functional reorganization. The hand tendon transfer is a rehabilitative surgery that restores the hand extension function by changing the insertion of chosen flexor muscles. During rehabilitation, the central control of the flexion and extension movements is based on elementary processes of brain plasticity, still poorly understood, but known to be slow, taking up to one year and occurring simultaneously with manual recovery. We hypothesize that high-resolution functional MRI (fMRI) may reveal brain plasticity linked to motor-behavior recovery.This PhD thesis aims to develop fMRI methods for individual patient follow-up after hand tendon transfer surgery. We first defined reference maps, based on a group of healthy subjects, for fine cortical representation of voluntary hand movements. These maps were then used to evaluate brain plasticity linked to the hand tendon surgery. To better understand the representation of the hand’s movements of flexion and extension in the primary motor cortex, we carried out a high-resolution, functional MRI study (using a static magnetic field of 3 Tesla), on a group of 13 healthy subjects. It was necessary to register all individual brains into a common reference system in order to perform a group statistical analysis. However, the inter-individual anatomical variability of the primary motor areas prevented us from performing the registration by conventional methods. We evaluated various methods of nonlinear registration to ensure a robust inter-individual alignment of central sulci. We compared: 1. a local method (Demon), which allows for important local deformations, 2. a standard non-linear method (SPM), which allows for a global image alignment and 3. a global diffeomorphic registration method (DARTEL) with a large number of degrees of freedom. In addition, we applied a fourth, more recent method, in partnership with LSIS, (CNRS Marseille) which iteratively reinforces the alignment of identified sulci before applying DARTEL (DISCO+DARTEL). This adds a local sulci-based constraint to the global deformation. We found that the fourth method produced the best alignment according to both anatomical criteria (Hausdorff distance, mask tissues overlaps and characteristics of the deformations field) and functional criteria (localization accuracy and statistical robustness of activations detected). We were then able to establish high spatial resolution reference maps of the elementary movements of the hand. These maps showed multiple foci of activity and significant overlaps in the region known as the "hand-knob."This first step of testing completed, we entered into a second round of testing where we evaluated the reproducibility of our initial results by performing a second series of tests on the same control group (“test-retest”).Two patients were followed up: before surgery (M0) and 1 month (M1), 3 (M3), 6 (M6) and 12 months (M12) after surgery. Functional recruitment was observed during recovery in SMA, M1, ispilateral and contralateral S1 and lobules V-VI of the cerebellum. Compared with healthy subjects, these two patients presented compensatory cortex reorganization during progressive recovery of hand function.

Durante um procedimento cirúrgico cerebral existe o deslocamento das estruturas que é um problema tipicamente não-rígido e não-linear. A ultrassonografia intra-operatória é utilizada como guia cirúrgico e pode ser utilizada para correção das imagens pré- operatórias através do corregistro rígido entre estas e um sistema de rastreio. Isto torna possível a visualização do deslocamento das estruturas devida a remoção de parte delas durante o ato cirúrgico. O objetivo deste trabalho é um estudo do corregistro livre não-rígido a partir de um modelo in vitro experimental que simule uma situação cirúrgica de retirada de uma inclusão líquida, de forma controlada, para medir os deslocamentos das estruturas próximas, utilizando imagens de ultrassom. Alguns fantomas que simulam o tecido humano nas imagens de ultrassom, feitos de gelatina e parafina, foram escolhidos como modelo. Para realizar o corregistro foi escolhida a transformação geométrica por splines simples (B-Splines), o otimizador Limited- memory BroydenFletcherGoldfarbShanno (LBFGS) e a métrica de similaridade soma do quadrado das diferenças (SQD) e, utilizada a biblioteca Insight Segmentation and Registration Toolkit (ITK), assim como o estudo dos parâmetros adequados para a nossa tarefa. Foi demonstrado para as condições envolvidas que para as imagens em modo B as deformações até 5% e mapas de RF até 9%, sem nenhuma otimização dos parâmetros do corregistro, é factível sem uso excessivo de tempo computacional. Foi analisada a influência da grade em relação a dois tipos diferentes de deformação, ambas com valor de 2%. O tamanho da grade, levando em consideração o erro e o tempo, foram a 5x11 para as imagens em Modo B e 11x17 para os mapas de RF, independentemente do tipo de deformação. Os parâmetros do otimizador (Default Step Length, Gradient Convergence Tolerance e Line Search Accuraccy) também foram avaliados e os valores obtidos foram 1,6; 0,03 e 0,8 para as imagens modo B e 1,2; 0,05 e 1,0 para os mapas de RF. No entanto ao comparamos, utilizando os parâmetros propostos obtidos, os campos de deslocamentos esperados com os gerados pelo modo B e pelos mapas RF, foi demonstrado que os mapas de RF fornecem valores abaixo do esperado e que as imagens em modo B retratam mais fielmente os deslocamentos e isto se deve a escolha do conjunto de valores testados para o otimizador. Foram aplicados estes parâmetros em dois fantomas de parafina- gel e em dois de gelatina. Nos três primeiros fantomas foi retirada um inclusão líquida em várias etapas. Os deslocamentos das estruturas vizinhas foram avaliados durante as etapas de remoção para demonstrar os campos de sução e de torção. No último fantoma, que simula morfologicamente um cérebro humano, foram retiradas, em várias etapas, regiões sólidas, simulando a retirada de tecido e foram calculados os deslocamentos e demonstrados os campos provenientes deste tipo de intervenção. Os trabalhos futuros se concentrarão em utilizar os volumes para medir os movimentos das estruturas e em novos parâmetros do otimizador para os mapas de RF. / During a brain surgery there is the displacement of the structures that is a typical non- rigid and non-linear problem. Intraoperative ultrasound is used as a surgical guide and can be used for spatial correction of preoperative images through the rigid registration between these and a track system. This makes it possible to visualize the displacement of structures due to removal of some piece of them during surgery. This work is a study of the non-rigid free-from registration using an experimental in vitro model to simulate a surgical situation withdrawal of a fluid inclusion in a controlled manner, to measure the displacement of nearby structures, using ultrasound images. Some phantoms that simulate the human tissue in the ultrasound images made of gelatin and paraffin were chosen as a model. To perform the registration it was used the framework Insight Segmentation and Registration Toolkit (ITK) and were chosen a geometric transformation of simple splines (B-splines), the Limited-memory Broyden-Fletcher- Goldfarb-Shanno (LBFGS) optimizer and the similarity metric sum of the squared differences (SQD). The search for the suitable parameters for our task are done and it has been shown that for the conditions involved for B-mode images deformations up to 5% and RF maps up to 9% without any optimization of the parameters of registration, is feasible without excessive use of computational time. The influence of the grid was examined for two different types of deformation, both for 2%. The size of the grid, taking into account the error and time were the 5x11 for the images in B mode and 11x17 maps for RF, regardless of the type of deformation. The parameters of the optimizer (Default Step Length, Gradient Convergence Tolerance and Line Search Accuraccy) were also evaluated and the values obtained were 1.6, 0.03 and 0.8 for the B-mode images and 1.2, 0.05 and 1.0 for RF maps. However when comparing the expected displacement fields with the generated by B-mode images and the RF maps, using the obtained parameters, it have been shown that RF maps provide values are lower than expected and that the B-mode images portray more faithfully displacements. This is due to the choice set of values tested for the optimizer. Finally, image registration parameters for B-mode were applied in two paraffin-gel and two gelatin phantoms. In the first three phantoms the fluid inclusion was removed in several stages and the displacements of neighboring structures were evaluated during the removal steps to demonstrate the fields of suction and torsion. The last phantom, which morphologically mimics a human brain, a solid region was removed, also in several stages, simulating a surgery. The displacements were calculated and demonstrated the fields from this type of intervention. Future work will focus on using the volumes to measure the movements of the structures and new parameters test of the optimizer to RF maps.

Abstract Automatic three-dimensional scene reconstruction from multiple images is a central problem in geometric computer vision. This thesis considers topics that are related to this problem area. New models and methods are presented for various tasks in such specific domains as camera calibration, image-based modeling and image matching. In particular, the main themes of the thesis are geometric camera calibration and quasi-dense image matching. In addition, a topic related to the estimation of two-view geometric relations is studied, namely, the computation of a planar homography from corresponding conics. Further, as an example of a reconstruction system, a structure-from-motion approach is presented for modeling sewer pipes from video sequences. In geometric camera calibration, the thesis concentrates on central cameras. A generic camera model and a plane-based camera calibration method are presented. The experiments with various real cameras show that the proposed calibration approach is applicable for conventional perspective cameras as well as for many omnidirectional cameras, such as fish-eye lens cameras. In addition, a method is presented for the self-calibration of radially symmetric central cameras from two-view point correspondences. In image matching, the thesis proposes a method for obtaining quasi-dense pixel matches between two wide baseline images. The method extends the match propagation algorithm to the wide baseline setting by using an affine model for the local geometric transformations between the images. Further, two adaptive propagation strategies are presented, where local texture properties are used for adjusting the local transformation estimates during the propagation. These extensions make the quasi-dense approach applicable for both rigid and non-rigid wide baseline matching. In this thesis, quasi-dense matching is additionally applied for piecewise image registration problems which are encountered in specific object recognition and motion segmentation. The proposed object recognition approach is based on grouping the quasi-dense matches between the model and test images into geometrically consistent groups, which are supposed to represent individual objects, whereafter the number and quality of grouped matches are used as recognition criteria. Finally, the proposed approach for dense two-view motion segmentation is built on a layer-based segmentation framework which utilizes grouped quasi-dense matches for initializing the motion layers, and is applicable under wide baseline conditions.

A common co-pathology of large lung tumors located near the central airways is collapse of portions of lung due to blockage of airflow by the tumor. Not only does the lung volume decrease as collapse occurs, but fluid from capillaries also fills the space no longer occupied by air, greatly altering tissue appearance. During radiotherapy, typically administered to the patient over multiple weeks, the tumor can dramatically shrink in response to the treatment, restoring airflow to the lung sections which were collapsed when therapy began. While return of normal lung function is a positive development, the change in anatomy presents problems for future radiation sessions since the treatment was planned on lung geometry which is no longer accurate. The treatment must be adapted to the new lung state so that the radiation continues to accurately target the tumor while safely avoiding healthy tissue. However, to account for the dose delivered previously, correspondences of anatomy between the former image when the lung was collapsed and the re-expanded lung in a current image must be obtained. This process, known as deformable image registration, is performed by registration software. Most registration algorithms assume that identical anatomy is contained in the images and that intensities of corresponding image elements are similar; both assumptions are untrue when collapsed lung re-expands. This work was to develop an algorithm which accurately registers images in the presence of lung expansion. The lung registration method matched CT images of patients aided by vessel enhancement and information of individual lobe boundaries. The algorithm was tested on eighteen patients with lung collapse using physician-specified correspondences to measure registration error. The image registration algorithm developed in this work which was designed for challenging lung patients resulted in accuracy comparable to that of other methods when large lung changes are absent.

京都大学 / 0048 / 新制・課程博士 / 博士(医学) / 甲第22642号 / 医博第4625号 / 新制||医||1044(附属図書館) / 京都大学大学院医学研究科医学専攻 / (主査)教授 武田 俊一, 教授 増永 慎一郎, 教授 鈴木 実 / 学位規則第4条第1項該当 / Doctor of Medical Science / Kyoto University / DFAM

Minimally invasive spine surgery (MISS) has proven to be advantageous over traditional open surgery as it minimizes the likelihood of tissue damage and infections. During MISS, surgeons create small incisions to allow access to the surgery site, however, opting for smaller incisions decreases the surgeon’s field of vision. To compensate, surgeons rely on preoperative and intraoperative ionizing imaging technologies for guidance. Conventional localization of the spine, registration of digital images to the patient during surgery, depends heavily on the surgeon’s anatomical knowledge and their experience. Preoperative images are typically created using 3D technology while intraoperative images use 2D technology. While the integration of preoperative 3D images and intraoperative 2D images can provide valuable assistance, patient’s preoperative and intraoperative positions do not coincide leading to additional use of ionizing imaging. The objective of this research was to propose a workflow that assists with image registration for MISS. The main component of the workflow was the creation of a script that automatically generates patient-specific digital guides, which will then be manufactured, to align the patient’s intraoperative and preoperative body position. By aligning the patient’s positions, the 3D printed surgical guide serves as a shared feature between the preoperative digital image and the actual patient. This allows for the intraoperative image to be registered to the preoperative image more accurately. Additionally, the guide acts as an attachment site for any additional instrument guides/supports. The surgical guide generating script utilizes the skin contour of patient’s torso region, extracted from medical images, to automatically produce the guide’s horizontal and vertical components. Adjustments are made to the components using CAD software before proceeding to manufacturing, via 3D printing, and assembly of the guide. To validate the workflow, more specifically the script’s ability to automatically generate surgical guides that fit over the patient’s back, a guide was created for a mannequin. The maximum gap between the mannequin and the horizontal components was 0.8 cm and 1.5 cm for the vertical component.

Image registration is the transformation of different sets of images into one coordinate system in order to align and overlay multiple images. Image registration is used in many fields such as medical imaging, remote sensing, and computer vision. It is very important in medical research, where multiple images are acquired from different sensors at various points in time. This allows doctors to monitor the effects of treatments on patients in a certain region of interest over time. In this thesis, artificial neural networks with curvelet keypoints are used to estimate the parameters of registration. Simulations show that the curvelet keypoints provide more accurate results than using the Discrete Cosine Transform (DCT) coefficients and Scale Invariant Feature Transform (SIFT) keypoints on rotation and scale parameter estimation.

This master’s thesis is aimed at registration of frames of retinal fundus video using phase corre- lation. An introduction describes general research in topic of retinal fundus, eye movements, diff erent approaches for image registration, phase correlation and examples of phase corre- lation applications. The second, practical part of master’s thesis, is dedicated to description of the proposed algorithm for registration of frames of retinal fundus video. The description of the proposed algorithm is divided into three parts. First two parts describe how frames of retinal fundus video are rated in terms of suitability for registration. Third part describes image registration algorithm itself. In conclusion, the accuracy of algorithm and computational time are evaluated.

This master thesis deals with retinal image registration using phase correlation technique. There are described properties of retinal images and modality of scanning. A geometrical transformation encompasing scale, rotation and translation between two retinal images is considered and the whole registration framework is described. There are used retinal images from fundus camera and scanning laser ophthalmoscope (SLO). In this thesis is described corresponding bifurcations detection using phase correlation and registration using second-order polynomial transformation. The results are subjectively and objectively verificated.

The aim of this thesis is to familiarize with the issue of registration of medical image sequences. The main objective was to focus on the method of feature tracking in the image and various options of its implementation. The theoretical part describes various methods for detection of feature points and future point matching methods. In the practical part these methods were implemented in Matlab programming environment and a simple graphical user interface was created.