Convective fluid motion through artery walls aids in transvascular transport of macromolecules. Although measurements of convective filtration have been reported, they were all obtained under constant transmural pressure. However, arterial pressure in vivo is pulsatile. Therefore experiments were designed to compare filtration under steady and pulsatile pressure conditions. Hydraulic conductance was measured in cannulated excised rabbit carotid arteries at steady pressure. Next, pulsatile pressure trains were applied within the same vessels, and simultaneously, arterial distension was monitored using Optical Coherence Tomography (OCT). For each pulse train, the volume of fluid lost through filtration was measured (subtracting volume change due to residual distension), and compared to that predicted from steady pressure measurements. In order to determine the role of the endothelium in this response, and the effect of increasing pulsatile frequency from an initial value, one of each pair was de-endothelialized in some cases, and in other experiments a pulsatile pressure of 1 Hz was initially applied, followed by a pulsatile frequency of 2 Hz. In all cases the experimental filtration volumes were significantly increased compared to those predicted for steady pressure, but over time, the magnitude of the excess fluid loss was reduced. For de-endothelialized vessels, this reduction was not so marked. These studies suggest that changes in arterial pulsatility may transiently increase convective flux of macromolecules into the artery wall and that this is regulated by the endothelium. In a parallel study, Bovine Aortic Endothelial Cells (BAEC) were exposed to a transient pressure gradient and then held at 20 mmHg for ten or thirty minutes. After staining for actin fibers and/or catenin, the cells were examined using a deconvolution microscope. The location of actin fibers changed from the body of the cell (central fibers) to the edges of the cell (peripheral fibers), and beta-catenin increased around the periphery. This result indicates that cultured endothelial cells can sense a change in transcellular pressures and respond so as to maintain cell-to-cell adhesion. Overall, the observed responses of arteries and endothelial cells to transient pressure gradients in these studies suggest a dynamic role for the endothelium in regulating transvascular transport in vivo.

In this dissertation, a methodology for comparing large arteries and tissue engineered vascular grafts is presented. This methodology is based on general porohyperelastic transport swelling theory (PHETS). Suites of experiments are introduced to determine material and transport properties of each vessel. These properties include elasticity, permeability, diffusivity, and convection coefficient. Finite element models (FEMs) were then used to model investigate arterial wall fluid flow and mobile species transport under quasi-static and pulsatile conditions. Rabbit carotid arteries were compared to rabbit aortas. The carotid was more elastic and permeable then the aorta. The pulsatile fluid wall flux was very different from the quasi-static and pulsatile in vivo conditions in these vessels. Tissue engineered vascular grafts (TEVGs) were fabricated in a bioreactor using high and low wall shear stress conditions. The elevated stiffness of ePTFE TEVGs significantly affects the fluid and species transport under both quasi-static and pulsatile conditions. A repeating influx/efflux condition developed in the large arteries and TEVGs during pulsatile pressurization. These conditions provide fluid/species transport pathways in arteries and TEVGs in pulsatile environments. The theoretical basis for ABAQUS FEMs coupled convection/diffusion of neutral species and water was developed. This will allow the analysis of mobile species concentration and flux in complex FEMs of soft biological structures. The theory and FEMs should also be useful in the study of vascular diseases, TEVG development, and drug transport in soft tissues.

The goal of this work was to assess the feasibility of using corona discharge quenching by anesthetic gases as a technique for anesthetic gas concentration measurement. Two experiments were conducted to investigate corona discharge and measure changes due to anesthetic gases. Experiment One used a chamber in which a high voltage was imposed across two parallel plane electrodes, between which gases under test could flow. Halothane, ethrane, and nitrous oxide were shown to have corona discharge quenching effects proportional to their relative potency. In an attempt to improve accuracy and decrease baseline drift a second system was fabricated. This system used an improved voltage source, temperature and humidity control and a chamber in which gases flowed between two concentric cylindrical electrodes. Results from the second experiment showed that the complex physics of corona discharge quenching by anesthetic gases could not be easily used for reliable measurement of anesthetic gas concentrations.

A system was developed for detecting gradual changes in background EEG, with a view to simplifying the tedious task of long-term monitoring in the neurological intensive care unit. / Six-channel EEG recordings serve as input to the system. EEG from each hemisphere is segmented into variable length epochs described by nine-dimensional feature vectors. A fixed scaling operation is performed so that differentiation between segments using the Euclidean distance between feature vectors relies equally on each feature. / Cluster analysis is performed to select five types of EEG that best represent the data spread for each hour. Color-coded circles representing clusters from every hour are mapped on a two-dimensional plot. The relative location of these circles reflects the difference between their associated EEG. Change between hours is detected by observing the movement of circles associated with each hour. / Detection of change with this method agreed closely with that observed via manual review but was much simpler.

One of the goals of current research in medical ultrasound is to develop techniques to quantitatively differentiate between tissue types and tissue states on the basis of the changes in the backscattered signal caused by differences in the elastic properties of the tissues under study. In this dissertation, we proposed to classify white matter and gray matter of adult brain tissues. The first step in this investigation was to develop a linear dichotomizer for the classification of white matter and gray matter using a power spectrum eigenvector approach with Linear Discriminant Analysis (LDA). The LDA classifier is described in Chapter 4. The successful classification rates for pure white matter and pure gray matter tissues were 99% using the LDA. When the LDA was used in transition regions the successful classification rates for both type of tissues dropped to 56%. We concluded that the weakness with this classification scheme was that it did not exploit the information at different scales. / In order to overcome this weakness, two multiscale classification methods were developed in Chapter 5, a power spectrum eigenvector approach with Wavelet transform-Linear Discriminant Analysis (WLDA) and a power spectrum eigenvector approach with Wavelet transform-Artificial Neural Network (WANN). Both methods were tested on calibrated media, also known as tissue mimicking phantoms, as well as on synthetically generated media. The data from synthetically generated media were obtained by a three Dimensional Finite Difference Time Domain method (3D FDTD). Both multiscale classification methods performed as well as the LDA method in classifying pure white matter and pure gray matter but they performed better than the LDA in transition regions. The successful classification rates in transition regions were 88% for white matter and 86% for gray matter using WLDA and 89% for white matter and 88% for gray matter using WANN.

We have developed and implemented an analytic simulation system to evaluate and correct quantitative imaging distortions in positron emission tomography (PET) scans. It is based on measured tomograph characteristics and realistic 3-D brain models generated from regionally segmented brain image data. Each structure is assigned with a regional radiotracer concentration and attenuation coefficient to create 3-D dynamic brain models and tissue attenuation maps. Projection data are then generated by incorporating key physical factors of detector geometry and resolution, attenuation, scatter, randoms, efficiency, deadtime and counting statistics. This has been done for a multi-slice PET scanner and includes temporal sampling and radiotracer decay. Simulated emission and transmission data are reconstructed by a filtered-backprojection algorithm. / The simulation methods are validated by scan data from both geometrical and anatomically realistic brain phantoms. Simulated projection components of a uniform phantom and a 3-D Hoffman brain phantom agree accurately with the measured data from our PET scanner. We then summarize current applications of this simulation tool to improve regional radioactivity quantification and optimize imaging protocols. In particular we have implemented a novel methodology to estimate and correct 3-D partial volume effects in dynamic PET studies using correlated magnetic resonance images. Simulations and phantom data in both single and double isotope experiments reveal substantial errors in striatal and cortical structures. Both show spatially variant and nonlinear distortions in time-activity curves which become more significant with degrading image resolution. These errors are removed completely by the partial volume correction algorithm with a reasonable increase in variance. / This software package is flexible and extensible. We have added many automatic steps to increase computational efficiency and simplify its usage in a clinical environment. This simulation tool offers a unified framework to evaluate and optimize PET imaging methodology from data acquisition, processing, and reconstruction to image analysis and physiological parameter estimation.

In this thesis, two new automatic image segmentation techniques are proposed and used to analyze medical magnetic resonance (MR) images of human brain medial temporal lobes. The first segmentation technique is an adaptive multi-contrast MR image based appearance modeling scheme, which combines level set and active appearance modeling methods and incorporates multi-contrast MR images, into segmentation. The contribution of each multi-contrast image to the segmentation is established by the correlation between each multi-contrast test gray image and its corresponding synthesized gray image. The latter is a linear combination of gray eigen-images characterized by principle component analysis as a result of converting a set of possibly correlated shape training images and multi-contrast gray training images into a set of linearly uncorrelated shape and multi-contrast gray eigen-images. Both the contributing weights and the linear combination model parameters are iteratively updated to minimize a weighted sum of least-square intensity differences between test image and synthesized image. The resulting model parameters are then used to linearly combine the shape related eigen-images to form a synthesized shape image as a final segmentation. In segmenting the hippocampi and amygdalae from MR images, this segmentation scheme with adaptive contributing weights is shown to provide better performance as measured by mean Dice κ values than its counterpart with equal contributing weights. The second segmentation technique is a two-stage segmentation approach, motivated by a concept of using local patches with similar intensity levels to make a collective decision so as to more accurately segment the structure boundary. In the two-stage segmentation, an appearance model-based global segmentation is employed as a first-stage segmentation to identify a coarse contour, while a patch-based local refinement is used as a second stage segmentation to make a local area correction, but only on a small set of voxels along the coarse contour identified earlier. In this thesis, the first stage segmentation uses only T1 images instead of multi-contrast images to avoid an increase in computational complexity. It is shown that the two-stage segmentation outperforms its one-stage counterpart in segmenting MR images of the human brain medial temporal lobe structures, including the hippocampus (HC), the amygdala (AG), the entorhinal/perirhinal cortex (EPC), and the parahippocampal cortex (PHC). Medial temporal lobe volumes, estimated by applying the two-stage segmentation on an MR database of 306 subjects with healthy brain development across a 4 to 18 year age range, are further analyzed and sex-specific growth patterns are derivedto help better understand puberty-related and sexually dimorphic brain maturation. Sexual-maturity level, measured by puberty scores, is used to partition the database into two groups: before and during puberty. The structure volumes for boys are largerthan those for girls but the difference varies between the AG, HC, EPC, and PHC. Age-related volumetric growth is observed in the left and right AG, left and right HC, right EPC, and left PHC and these volumetric changes are statistically significant, but only before puberty. After onset of puberty, volumetric growth tends to correlate with sexual maturity level. When evaluated with head size normalized volumes, we find smaller volumes of the right HC, left and right PHC, for more sexually matureboys, and larger volumes of the left HC for more sexually mature girls. These findings suggest that the rising levels of testosterone in boys and estrogen in girls might have opposite effects, especially for the HC and the PHC. Our findings on sex-specific and sexual maturity-related volumes may be useful in better understanding the medial temporal lobe developmental differences and related learning, memory, and emotion differences between boys and girls during puberty. / Dans ce travail de thèse, deux nouvelles méthodes de segmentation automatique d'Image par Résonance Magnétique (IRM) sont proposées. Ces deux méthodes sont appliquées et validées dans le cadre de la segmentation des structures du lobe medio temporal du cerveau humain. La première méthode de segmentation proposée repose sur un modèle d'apparence de forme adaptatif utilisant simultanément plusieurs séquences d'IRM. Cette méthode combine des ensembles de niveaux et une modélisation d'apparence de formes actives tout en incorporant l'information issue de différents contrastes en IRM. L'importance de chaque séquence IRM est estimée automatiquement au sein d'un schéma adaptatif en utilisant la corrélation des différents contrastes avec les images synthétisées correspondantes. Ces images synthétisées sont construites par combinaison linéaire des images propres obtenues par Analyse en Composante Principale (ACP). Les paramètres du model de forme ainsi que les poids attribués à chaque contraste sont ensuite estimés itérativement via une minimisation pondérée des moindres carrés entre l'image test et l'image synthétisée. Finalement, les paramètres obtenus sont utilisés pour combiner les formes propres afin d'obtenir la segmentation finale. La méthode proposée est validée dans le contexte de la segmentation de deux structures cérébrales: l'hippocampe et l'amygdale. Lors de cette validation, l'amélioration de la qualité de segmentation apportée pour le schéma adaptatif, comparé à utilisation de poids identiques pour chaque contraste, est étudiée en utilisant l'index Kappa comme mesure de qualité. La deuxième méthode de segmentation proposée combine deux étapes: une initialisation globale par un modèle d'apparence de forme puis un raffinement local utilisant une approche par patches. Cette seconde étape tire avantage des performancesdes méthodes de type moyennes non locales qui estiment la similarité de deux échantillons en calculant la distance de niveau de gris de leur voisinage sous forme de patches. Dans l'approche proposée, le modèle d'apparence est destiné à obtenir rapidement une première estimation de la segmentation utilisée comme initialisation. Ensuite, le raffinement local par patches est appliqué uniquement sur un petit ensemble de voxels le long des bords de la segmentation initiale. Une validation est effectuée sur plusieurs structures du lobe medio temporal. Cette validation démontre l'intérêt de combiner les deux étapes décrites comparé à l'utilisation non conjointe de chacune des méthodes. Une étude volumétrique des structures du lobe medio temporale est menée à l'aide de la seconde méthode sur 306 sujets sains âgés de 4 à 18 ans. L'influence du sexe des sujets sur la croissance des structures est étudiée afin de permettre une meilleure compréhension des différences de maturation cérébrale et des phénomènes liés à la puberté. Le niveau de maturité sexuelle, mesuré à l'aide du score de la puberté, est utilisé afin de séparer la base de données en deux groupes: avant et pendant la puberté. Des changements liés à l'âge sont observés pour l'AG droite et gauche, l'HC droit et gauche, le CEP droit and le CPHC gauche. Ces changements sont statistiquement significatifs mais seulement avant la puberté. Pendant la puberté, l'augmentation du volume tend à être corrélé avec le niveau de maturité sexuel. En utilisant les volumes normalisés par la taille de la tête, des volumes plus faibles de l'HC droit et du CPHC droit et gauche sont observé pour les garçons les plus matures sexuellement et des volumes plus grands pour l'HC gauche sont obtenus pour les filles les plus matures sexuellement. Ces résultats sur la croissance des structures du lobe medio temporal selon le sexe et la maturité peuvent participer à l'amélioration de notre compréhension des différences observées entre le filles et le garçons durant la puberté au niveau de l'apprentissage, de la mémoire et des émotions.

In order to claim that the behaviours of early visual cortex neurons are fully understood, it must be possible to create models of them that can predict responses to any arbitrary type of stimuli. Thus the underlying theme of this thesis is to explore stimuli and implement system identification approaches appropriate for the estimation of receptive field (RF) models, as well as to quantitatively examine their response properties. Specifically, I aim to address the following questions: 1) what types of stimuli provide the most robust and generalizable RF models? 2) are cortical simple cell RF models classified into distinct categories? and 3) what are the spatiotemporal response properties of the RF models?To answer these questions, I use artificial and natural stimuli in conjunction with regularized regression system identification, to estimate RF models of visual cortical neurons. The use of natural stimuli is more appropriate as they drive more ecologically relevant responses. Regularized regression system identification techniques, which have antecedents in the field of machine learning, have a number of advantages: 1) they are based on regression and therefore appropriate for a wide range of stimuli including natural images; 2) they incorporate regularization to prevent overfitting; and 3) they allow for the estimation of full 3D (i.e., space-space-time) spatiotemporal RFs in a computationally tractable manner. I investigate simple cells of a somewhat higher cortical processing area (cat area 18), thereby providing a more challenging medium for our experiments. Results from Chapter 2 showed that white noise stimuli usually elicited weak responses, yielding poor RF models, while short bar and natural image stimuli drove strong responses yielding RF models with high predictive ability. However, RF models derived from natural images better predicted responses to other types of novel stimuli, when compared to models derived from broadband artificial stimuli (i.e., white noise, short bars). Furthermore, natural image-derived RF models performed well when used to predict responses to sinewave gratings.Chapter 3 revealed that early cortical simple cells could be delineated into three distinct categories, based on spatial RF structure and output nonlinearity. These three classes were: 1) non-oriented cells with expansive nonlinearities; 2) oriented cells with expansive nonlinearities; and 3) oriented cells with compressive nonlinearities. Our sample was split between these three types, suggesting that large numbers of non-oriented cells could be readily found in brain areas beyond the lateral geniculate nucleus and layer 4 striate cortex. Additionally, our results suggested that compressive output nonlinearities were much more common than previously believed. In Chapter 4 I identified a number of spatiotemporal response properties that varied amongst the three types of simple cells. These included optimal spatial frequency, latency, duration, responsivity, and direction selectivity. Of these, optimal spatial frequency was the most interesting as it showed segregation of our sample into distinct clusters, suggesting that optimal spatial frequency may be another dimension along which categorically distinct classes of cortical neurons could be established. Results also revealed that the three cell types effectively covered a large range of temporal latencies and durations. Finally, the two classes of oriented cells differed in their levels of responsivity, suggesting that expansive versus compressive nonlinearities may reflect the degree to which a nonlinear gain control mechanism is engaged. Collectively, the three studies in this thesis demonstrate the powerful potential of natural image stimuli and regularized regression system identification techniques, making them ideal candidates for routine use in visual neuroscience studies. RF models derived from natural images have revealed that physiological response... / Afin de prétendre que les comportements des neurones du cortex visuel sont pleinement compris, il doit être possible de créer des modèles de ceux-ci qui peuvent entièrement prédire leurs réponses à tout type arbitraire de stimuli. Le thème sous-jacent de cette thèse est donc d'explorer les stimuli et les approches d'identification de système appropriés pour l'estimation de tels modèles de champ récepteur (CR), ainsi que d'examiner de façon quantitative les propriétés de leur réponse. Plus précisément, je cherche à répondre aux questions suivantes: 1) quels types de stimuli donnent des modèles de CR les plus robustes et généralisables?; 2) est-ce que les modèles de CR des cellules corticales simples sont classés en catégories distinctes?; et 3) quelles sont les propriétés des réponses spatiotemporelles des modèles de CR?Pour répondre à ces questions, j'utilise des stimuli artificiels et naturels en conjonction avec l'identification de système par régression régularisée, pour estimer les modèles de CR des neurones du cortex visuel. L'utilisation de stimuli naturels est plus appropriée, car elle entraîne des réponses plus écologiquement pertinentes. Les techniques d'identification de système par régression régularisée, qui ont des antécédents dans le domaine de l'apprentissage automatique, comportent un certain nombre d'avantages: 1) elles sont basées sur la régression et donc appropriées pour un large éventail de stimuli, y compris les images naturelles; 2) elles incorporent la régularisation afin d'éviter le sur-ajustement; et 3) elles permettent l'estimation de CR spatiotemporels 3D complets (c.-à-d., espace-espace-temps) de manière calculable par ordinateur. J'étudie les cellules simples d'une région corticale de niveau de traitement légèrement plus haut (chat, aire 18), offrant ainsi un milieu plus stimulant pour nos expériences. Les résultats du Chapitre 2 ont démontré que l'utilisation de bruit blanc comme stimuli suscitait généralement de faibles réponses, produisant de pauvres modèles de CR, tandis que l'utilisation de barres courtes et d'images naturelles comme stimuli provoquait de fortes réponses, ce qui donne des modèles de CR ayant une grande capacité prédictive. Toutefois, les modèles de CR dérivés d'images naturelles ont mieux prédit les réponses à d'autres types de nouveaux stimuli, par rapport aux modèles dérivés de stimuli artificiels à large bande (c.-à-d., bruit blanc, barres courtes). De plus, les modèles de CR dérivés d'images naturelles ont bien performé lorsqu'utilisés pour prédire les réponses aux grilles sinusoïdales. Le Chapitre 3 a révélé que les cellules corticales simples pouvaient être délimitées en trois catégories distinctes, basées sur la structure spatiale des CR et sur la non-linéarité de sortie. Ces trois catégories étaient: 1) les cellules non-orientées avec des non-linéarités à expansion; 2) les cellules orientées avec des non-linéarités à expansion; et 3) les cellules orientées avec des non-linéarités à compression. Notre échantillon a été réparti entre ces trois types, ce qui suggère qu'un grand nombre de cellules non-orientées peuvent être facilement trouvées dans les zones du cerveau au-delà du corps genouillé latéral et de la couche 4 du cortex strié. De plus, nos résultats suggèrent que les non-linéarités à compression de sortie sont beaucoup plus fréquentes qu'on ne le croyait auparavant. Dans le Chapitre 4, j'ai identifié un certain nombre de propriétés des réponses spatiotemporelles qui variaient entre les trois types de cellules simples. Il s'agissait notamment de la fréquence spatiale optimale, de la latence, de la durée, de la réactivité et de la sélectivité directionnelle. Parmi celles-ci, la fréquence spatiale optimale était la plus intéressante, car elle a démontré la ségrégation de notre échantillon en grappes distinctes. Ceci suggère que la fréquence spatiale optimale peut être une autre dimension au long de laquelle des classes de neurones corticaux...

Accurate surgical navigation for pedicle screw implantation significantly reduces the complication rate in spinal fusion surgeries. To achieve accurate navigation based on preoperative computed tomography (CT), accurate patient-to-image registration is necessary. However, current manual registration techniques are invasive and greatly prolong operative time. To resolve these shortcomings, automated patient-to-image registration through tracked intraoperative ultrasound has been proposed, but existing ultrasound-CT registration techniques are limited and not ready for clinical application in spine surgeries. This thesis presents the development of a technique for ultrasound-CT image registration of vertebrae that satisfies the practical requirements of being automated, accurate, robust, reasonably fast and appropriately validated. The ultrasound-CT registration technique first extracts the vertebral bone surface from both the ultrasound and CT images through scan line tracing. The extracted surfaces are then registered by intensity cross-correlation. Preliminary registration results on a single vertebra of a plastic Sawbones phantom yielded target registration error (TRE) under 1 mm. Subsequently, the technique was extensively validated on 18 vertebrae of 3 porcine cadavers, with a total of 18,000 registrations. All validation was with respect to gold standard registrations generated with imaging fiducials. The results demonstrated good registration accuracy, with median TRE of 1.65 mm, and good robustness, with 82.7% of TREs > 2 mm. The generated gold standard registration had a median TRE of 0.718 mm. The registration technique was further improved by eliminating the step of reconstructing ultrasound image slices into a volume. This was achieved by directly registering the ultrasound slices as a group to the target CT image volume. This improvement significantly reduced the total registration time from 8 min down to 4 min. The registration accuracy and robustness were also slightly improved, with a median TRE of 1.45 mm and 84.6% of TREs > 2 mm. In addition, a trade-off between registration accuracy and speed was established through the number of ultrasound image slices used in the registration.The technique of ultrasound-CT registration of vertebrae developed in this thesis is automated, accurate, robust, quick and practical for intraoperative use. In the future, validation on human cadavers and patients will enable the technique to be applied in clinical settings. / La navigation chirurgicale précise du vissage p′ediculaire réduit de façon significative le taux de complication des chirurgies de fusion spinale. La précision de la navigation basée sur la tomographie axiale (CT) préopératoire exige l'exactitude du recalage patient-image. Actuellement, les techniques de recalage manuelles sont invasives et prolongent le temps opératoire. Le recalage patient-image automatique par l'intermédiaire d'échographie peropératoire localisé a été proposé, mais les méthodes existantes de recalage d'échographie-CT sont limitées et ne sont pas prées pour l'usage clinique dans les chirurgies spinales. Cette thèse présente le dévelopement d'une technique pour le recalage échographie-CT vertébrale. Cette technique satisfait aux exigences pragmatiques d'être automatique, précise, robuste, avoir une vitesse raisonnable et d'être validée d'une manière adéquate.La technique de recalage d'échographie-CT extrait tout d'abord la surface osseuse de vertèbre à partir des images échographiques et CT. Ensuite, les surfaces extraites sont recalée par la corrélation croisée d'intensité de voxels. Le recalage d'un seul vertèbre d'un fantôme Sawbones en plastique a donné comme résultat préliminaire une erreur de recalage de cible ("target registration error" ou TRE) au-dessous de 1 mm. Par la suite, la technique a été validée plus largement sur 18 vertèbre de 3 cadavres porcins, un total de 18,000 recalages. Les validations ont employé les recalages d'étalon d'or générés avec des points de repères d'imagerie. Les résultats démontrent de bonnes précisions de recalage, ayant un TRE médian de 1.65 mm, et de bonne robustesse avec 82.7% des TREs > 2 mm. Le recalage d'étalon d'or avait un TRE médian de 0.718 mm.La technique de recalage a été amélioré en eliminant l'étape de la reconstruction des secteurs d'image d'échographie en un volume. Ceci a été accompli en recalant directement les secteurs d'échographie comme un group au volume d'image cible CT. Cette amélioration a réduit le temps total de recalage de 8 min à 4 min. La précision et robustesse étaient aussi améliorés légèrement, avec un TRE médian de 1.45 mm et 84.6% de TREs > 2 mm. De plus, un compromis entre la précision et la vitesse de recalage a été établi par le nombre de secteurs d'échographie utilisés dans le recalage.Cette thèse presente le developement d'une technique de recalage échographie-CT vertébrale automatique, précise, robuste, rapide et pratique pour l'usage peropératoire. Dans l'avenir, la validation sur les cadavres humains et les patients vont permettre l'adoption de cette technique à l'usage clinique.

The vocal folds are two membranous lips of soft tissue located within the human larynx. During phonation, they undergo self-sustained oscillations. Common voice disorders are believed to result from excessive mechanical stresses within the vocal folds' mucosal layer. The overarching goal of the present study was to better understand the relationship between mechanical loading and vocal fold tissue response. The bulk mechanical properties of the vocal folds were initially investigated using uniaxial traction testing and shear rheometry. These methods were used to quantify the material parameters of porcine vocal folds. A linear transversely-isotropic model was used to relate stresses and strains. The assumption of incompressibility was used to reduce the number of independent parameters.The effects of water loss induced by an osmotic pressure potential on vocal fold tissue's viscoelastic properties were investigated. Uniaxial traction testing was used to impose slow-rate, cyclical extensions of porcine vocal folds while a hypertonic solution was used to absorb interstitial fluid from the tissue. The elastic modulus and the loss factor were then determined for normal and dehydrated tissues. A non-linear eight-chain hyperelastic model was used to relate the stress and the stretch of the bi-phasic tissue. Significant changes in mass were observed as a result of the traction tests. The mechanics of the vocal folds were investigated under linear poroelasticity, where interstitial fluids are assumed to be freely moving within the extracellular matrix proteins. The one-dimensional consolidation problem was used to model the contact between tissue and a spherical indenter. Atomic force microscopy based indentation data, obtained from creep and dynamic oscillation testing, were used to calibrate the model. Nanoscale viscoelastic characteristics of porcine lamina propria were characterized from the force response to frequency-dependent displacement oscillations, with an amplitude of 30-50 nm. Nonlinear laser scanning microscopy was used to visualize the morphology of extracellular matrix proteins within human and animal vocal folds. A custom-built multimodal nonlinear laser scanning microscope was used to scan fibrous proteins in human and porcine vocal folds. Collagen and elastin were imaged using second-harmonic generation and two-photon fluorescence, respectively. An experimental protocol was introduced to characterize the geometrical properties of the collagen fibrils. Nonlinear laser scanning microscopy was then used to investigate the remodeling of scarred rat vocal folds. The volume fraction of collagen was found to be 12% greater in scarred vocal fold tissue 12 month after injury. Atomic force microscopy images suggest a rope-shaped structure of collagen fibrils in the vocal folds. A hyperelastic theory was developed for collagen-reinforced soft tissues. The relevant formulation was derived for finite element simulations. The model captured the role of helical hierarchies of the collagen fibrils in the nonlinear response of vocal folds under load. / Les cordes vocales sont des membranes de tissus mous situées à l'intérieur du larynx. Pendant la phonation, elles sont soumises à des oscillations auto-entretenues. Certains troubles de la voix répandus sont connus pour résulter de contraintes mécaniques excessives au sein de la muqueuse des cordes vocales. Les propriétés viscoélastiques des cordes vocales pathologiques diffèrent de celles dont les tissus sont sains. L'objectif global de cette étude est de mieux comprendre la relation entre le chargement mécanique et la réponse des tissus des cordes vocales. Les propriétés mécaniques de compression des cordes vocales ont tout d'abord été étudiées à l'aide d'essai de traction et d'un rhéomètre à cisaillement. Ces méthodes ont servi à quantifier les paramètres mécaniques de cordes vocales porcines. Un modèle linéaire, isotrope transverse a été utilisé pour la relation entre les contraintes et les déformations. La condition d'incompressibilité a permis de réduire le nombre de paramètres indépendants. Les effets de déshydratation, induite par le potentiel de pression osmotique, sur les propriétés des tissus des cordes vocales ont été étudiés. Des essais de traction uniaxiaux ont servi pour imposer des extensions cycliques à faible vitesse sur des cordes vocales porcines pendant qu'une solution hypertonique absorbait le fluide interstitiel des tissus. Le module élastique et le facteur de perte ont été calculés pour des tissus normaux et déshydratés. Un modèle d'hyperélasticité non linéaire à huit chaînes a servi pour décrire la relation entre les contraintes et les déformations du tissu biphasique. Des variations de masse significatives ont été observées à la suite des essais de traction. La mécanique des cordes vocales a été étudiée à l'aide de conditions de poroélasticité linéaire. Les fluides interstitiels sont supposés libres de mouvement au sein des matrices extra-cellulaires des protéines. Le problème de consolidation à une dimension a servi à la modélisation du contact entre les tissus mous et une indentation sphérique. Les données d'entrée du modèle étaient obtenues par la microscopie à force atomique basée sur des données d'indentation utilisant des signaux de rampe ou d'oscillations dynamiques. Des caractéristiques viscoélastiques furent mises en valeur à partir de la réponse des cordes vocales aux oscillations, dont le déplacement était contrôlé en fréquence, avec une amplitude de 30 à 50 nm. La microscopie optique non linéaire a permis la visualisation de la morphologie des matrices extra-cellulaires des protéines au sein de cordes vocales humaines et porcines. Un microscope non linéaire multimodale a été conçu pour scanner les protéines fibreuses de cordes vocales humaines et porcines. Le collagène et l'élastine ont été imagés respectivement à l'aide de la génération de second harmonique et de la fluorescence sous excitation à deux photons. L'introduction d'un protocole expérimental a servi à caractériser les propriétés géométriques des fibres de collagène. Cette méthode d'imagerie a ensuite été utilisée pour étudier le remodelage de cordes vocales de rats cicatrisées. Ceci a permis de montrer que la fraction volumique de collagène était 12% plus importante dans les tissus de cordes vocales cicatrisées 12 mois après la blessure.Les images du microscope à force atomique suggèrent que des fibres de collagène avec une structure de corde sont présentes dans les cordes vocales. Une théorie hyperélastique a été développée pour des tissus mous supportés par le collagène, ainsi que la formulation adaptée pour les calculs éléments finis. Le modèle capture le rôle de la structure hélicoïdale des fibres de collagène d'après la réponse non linéaire des cordes vocales soumises à un chargement.