La technique de spectroscopie basée sur la diffusion Raman Stokes spontanée est un procédé standard employé dans de nombreux domaines allant de la thermodynamique à la médecine, en passant par la science des matériaux. À la faveur d'un échange d'énergie inélastique, elle permet de déterminer les fréquences des vibrations moléculaires présentes dans un objet. On peut ainsi remonter à l'identification des molécules et ainsi caractériser l'objet d'étude sans utiliser de marqueur spécifique. Cette méthode est néanmoins affligée de défauts. Outre la présence d'un signal de fluorescence qui peut submerger la réponse Raman, le désavantage majeur est le long temps d'exposition que requière cette technique. Dans le cas d'étude d'échantillon biologique, cela proscris son usage pour des mesures de microspectroscopie : la cartographie spectrale d'objet microscopique. Afin de pallier ce problème, de nouvelles techniques ont été développées. C'est le cas de la spectroscopie employant la diffusion Raman anti-Stokes Cohérente (ou CARS pour Coherent Anti-Stokes Raman Scattering). Du fait de sa cohérence et de sa directivité le signal anti-Stokes affiche une intensité 10^5 to 10^6 fois plus importante que dans le cas de la diffusion Raman spontanée, ce qui permet alors d'abaisser le temps d'exposition à un niveau tolérable pour les objets biologiques lors d'une mesure de microspectroscopie. De plus, le caractère anti-Stokes du signal l'épargne de la contribution de la fluorescence. Pourtant, un défaut majeur limite encore l'utilisation de cette technique : le bruit de fond non résonant. Ce phénomène peut diminuer, voir noyer la contribution résonante qui porte l'information. Cette thèse a permis le développement de techniques CARS autorisant une réduction du bruit de fond non résonant. Pour ce faire un dispositif de spectroscopie CARS multiplex (M-CARS) en configuration copropagative a été construit. Ses capacités sont illustrées par des mesures spectrales d'échantillons minéral, végétal et biologique. À partir de ce système, il a été établi une méthode innovante permettant de discriminer le signal résonant du bruit non résonant en utilisant un champ électrique continu. Il est aussi démontré la mise en place d'un procédé qui a permis de mener la première mesure de microspectroscopie M-CARS en configuration contrapropagative sur un échantillon biologique. Cette configuration limite la collecte du signal à l'objet d'étude, empêchant ainsi l'acquisition du signal résonant et non résonant issu du solvant, principal responsable du bruit de fond non résonant lors d'une mesure CARS en configuration copropagative. / The spectroscopy technique based on spontanée Raman Stokes scattering is a standard process used in many fields spanning from thermodynamic and medicine, to materials sciences. An inelastic energy exchange permits to determinate the frequency of the molecular vibrations in an object. One can identify the molecules and thus, can characterize the object of study in a label-free way. Nevertheless, this method is afflicted with faults. Beside the presence of fluorecence that can drown the Raman answer, the main drawback is the long exposition time required. In the case of biological sample, this can prohibit the use of spontaneous Raman scattering for microspectroscopy measures: the spectral mapping of microscopic objects. To avoid this problem, new techniques have been developed. It is the case of Coherent anti-Stokes Raman scattering (CARS) spectroscopy. Due to its coherence and its directivity, the anti-Stokes signal has an intensity 105 to 106 times greater than the spontaneous Raman scattering one. The exposition time is then reduced to a tolerable level for biological objects during microspectroscopy measures. Moreover, the anti-Stokes characteristic of the signal prevents the fluorescence contribution. However, a major fault still limits the use of this technique: the nonresonant background. This phenomenon can diminish, even overwhelm the resonant contribution carrying the information. This thesis permitted the development of CARS approaches that allow the reduction of the nonresonant background. To do so, a multiplex CARS (M-CARS) spectroscopy apparatus in a forward configuration has been built. Its abilities are illustrated with spectral measures of mineral, vegetal and biological samples. Based on this system, it has been established an innovative method that can discriminate the resonant signal from the nonresonant one thanks to a static electric field. It has been also been demonstrated the development of a process that has allowed the first M-CARS microspectroscopy measure of a biological sample in a contrapropagative configuration. This setup limits the collect of the signal to the object of study, avoiding the acquisition of the resonant and resonant signals coming from the solvent, responsible for the major part of non resonant background during a CARS measure in a forward configuration.

No description available.

<p>Magnetic resonance spectroscopy (MRS) is a non-invasive diagnostic technique that provides unique information about the biochemical composition of the human body. By excluding the overwhelming signals from water and fat, clinically relevant biomarkers such as lactate, N-acetyl aspartate, choline, creatine, glutamate/glutamine (Glx), gamma-aminobutyric acid (GABA), glutathione, and myoinositol can be reliably quantified. MRS has diverse applications in investigating the metabolic window of a wide range of biochemical processes. </p> <p>Here, we have utilized MRS to better understand chemical changes associated with neurological disorders and treatment response. We have investigated neurometabolic imbalances in brain regions related to post-traumatic stress disorder (PTSD). MRS was applied to better understand the neurobiological processes of hyperbaric oxygen therapy in military veterans with clinically diagnosed traumatic brain injury and/or PTSD.</p>

<p>  </p> <p>In this study, we present advancements in smart contact lenses, highlighting their potential as minimally or non-invasive diagnostic and drug delivery platforms. The eyes, rich in physiological and diagnostic data, make contact lens sensors an effective tool for disease diagnosis. These sensors, particularly smart contact lenses, can measure various biomolecules like glucose, urea, ascorbate, and electrolytes (Na+, K+, Cl-, HCO3-) in ocular fluids, along with physical biomarkers such as movement of the eye, intraocular pressure (IOP) and ocular surface temperature (OST).</p> <p>The study explores the use of continuous, non-invasive contact lens sensors in clinical or point-of-care settings. Although promising, their practical application is hindered by the developmental stage of the field. This thesis addresses these challenges by examining the integration of contact lens sensors, covering their working principle, fabrication, sensitivity, and readout mechanisms, with a focus on monitoring glaucoma and eye health conditions like dry eye syndrome and inflammation.</p> <p>Our design adapts these sensors to fit various corneal curvatures and thicknesses. The lenses can visually indicate IOP through microfluidic channels' mechanical deformation under ambulatory conditions. We also introduce a colorimetric hydrogel tear fluid sensor that detects pH, electrolytes, and ocular surface temperature, indicating conditions like dry eye disease and inflammation.</p> <p>The evaluation of these contact lens sensors includes in vivo/vitro biocompatibility, ex vivo functionality studies, and in vivo safety assessments. Our comprehensive analysis aims to enhance the practicality and effectiveness of smart contact lenses in ophthalmic diagnostics and therapeutics.</p>

<p>  </p> <p>Heart disease is the leading cause of death for individuals with Duchenne muscular dystrophy (DMD). DMD is a devastating and progressive neuromuscular disease with no known cure. This X-linked genetic disorder affects nearly 1 in 5000 boys and manifests as debilitating muscle weakness and progressive cardiomyopathy (CM). While CM in some individuals with DMD progresses rapidly and fatally in their teenage years, others can live relatively symptom-free into their thirties or forties. Early identification and treatment can improve quality and length of life, but currently, there are no standard imaging biomarkers that can detect early onset or rapidly progressing DMD CM. Addressing this gap, we describe here a novel cardiac image analysis paradigm using 4D cardiac magnetic resonance imaging (CMR) to map left-ventricular kinematics comprehensively in DMD CM. The primary goal of this dissertation work is to introduce novel imaging biomarkers and computational methods to enable earlier diagnosis and precise prognosis for cardiac function in DMD. Central to this goal, we identified myocardial strain biomarkers that predict the early onset and rapid progression of cardiac disease in vulnerable patients. These findings bridge clinical gaps and pave the way for multi-center studies to characterize DMD CM progression and assessment of individual patient risk profiles for improved treatment and outcomes in DMD.</p>

Computed tomography has been widely used in medical diagnosis to generate accurate images of the body's internal organs. However, cancer risk is associated with high X-ray dose CT scans, limiting its applicability in medical diagnosis and telemedicine applications. CT scans acquired at low X-ray dose generate low-quality images with noise and streaking artifacts. Therefore we develop a deep learning-based CT image enhancement algorithm for improving the quality of low-dose CT images. Our algorithm uses a convolution neural network called DenseNet and Deconvolution network (DDnet) to remove noise and artifacts from the input image. To evaluate its advantages in medical diagnosis, we use DDnet to enhance chest CT scans of COVID-19 patients. We show that image enhancement can improve the accuracy of COVID-19 diagnosis (~5% improvement), using a framework consisting of AI-based tools. For training and inference of the image enhancement AI model, we use heterogeneous computing platform for accelerating the execution and decreasing the turnaround time. Specifically, we use multiple GPUs in distributed setup to exploit batch-level parallelism during training. We achieve approximately 7x speedup with 8 GPUs running in parallel compared to training DDnet on a single GPU. For inference, we implement DDnet using OpenCL and evaluate its performance on multi-core CPU, many-core GPU, and FPGA. Our OpenCL implementation is at least 2x faster than analogous PyTorch implementation on each platform and achieves comparable performance between CPU and FPGA, while FPGA operated at a much lower frequency. / Master of Science / Computed tomography has been widely used in the medical diagnosis of diseases, such as cancer/tumor, viral pneumonia, and more recently, COVID-19. However, the risk of cancer associated with X-ray dose in CT scans limits the use of computed tomography in biomedical imaging. Therefore we develop a deep learning-based image enhancement algorithm that can be used with low X-ray dose computed tomography scanners to generate high-quality CT images. The algorithm uses a state-of-the-art convolution neural network for increased performance and computational efficiency. Further, we use image enhancement algorithm to develop a framework of AI-based tools to improve the accuracy of COVID-19 diagnosis. We test and validate the framework with clinical COVID-19 data. Our framework applies to the diagnosis of COVID-19 and its variants, and other diseases that can be diagnosed via computed tomography. We utilize high-performance computing techniques to reduce the execution time of training and testing AI models in our framework. We also evaluate the efficacy of training and inference of the neural network on heterogeneous computing platforms, including multi-core CPU, many-core GPU, and field-programmable gate arrays (FPGA), in terms of speed and power consumption.

<p dir="ltr">Type 2 diabetes (T2D), a multi-factorial disease characterized by insulin resistance and pancreatic β cell dysfunction, accounts for 90% of all forms of diabetes. Recent studies identified β cell dedifferentiation as one mechanism leading to β cell failure in T2D, and this phenotype has been linked with loss of expression of cellular identity markers​. The molecular mechanisms leading to β cell dedifferentiation in T2D are not well defined and represent a key gap in our understanding of disease progression. Furthermore, sex-specific mechanisms contributing to β cell dysfunction remain unknown.<i> </i>We demonstrated that β cell-specific stromal interaction molecule 1 knockout (STIM1Δβ) leads to obesity-induced glucose intolerance and reduced expression of β cell identity genes in female but not male mice. Mechanistic studies identified reduced G-protein coupled estrogen receptor (GPER) expression in female STIM1Δβ islets, suggesting a novel connection between STIM1, GPER, and β cell identity. To determine how STIM1 and GPER interact to maintain β cell identity in females, we generated and characterized two mouse models: a surgical ovariectomy (OVX) model and a β cell-specific GPER knockout (GPERΔβ) model. Female high fat diet (HFD)-fed OVX mice showed glucose intolerance and loss of β cell identity gene expression, and <i>ex vivo</i> treatment of OVX islets with a GPER agonist rescued the loss of β cell identity. Additionally, HFD-fed GPERΔβ female mice had increased glucose excursions compared to controls. Our results highlight the critical role that estradiol signaling through GPER plays in the maintenance of β cell health and identity and the pathogenesis of T2D in females.</p>

<p dir="ltr">Chronic venous obstruction (CVO) is a prevalent but often underdiagnosed condition affecting the deep veins of the lower extremities. CVO arises from venous stasis caused by compression, valve damage, proximal vein disease, or post-thrombotic changes. This leads to the formation of collagen-rich lesions that impair blood flow, thicken vessel walls, and elevate pressure. While arterial remodeling is well studied, the impact of altered flow in the venous system due to disease remains less understood. This study aims to address the gap in knowledge of venous flow conditions by developing clinically relevant benchtop models to investigate the effects of variable inflow conditions in CVO. Benchtop models were constructed using polydimethylsiloxane and a silicone rubber, with the iliofemoral vein as the target (diseased) segment, the popliteal, great saphenous, and deep femoral as the inflow veins, and the common iliac vein as the outflow segment. One healthy model and nine diseased models were developed with varying sizes of CVO lesions in the iliofemoral vein. Ultrasound studies were conducted in each of the models to assess the impact of different inflow conditions on velocities within and around the lesion. In one model, particle-tracking was used to visualize flow patterns around the lesion. Doppler ultrasound measurements showed that as the cross-sectional area and length of the lesion increased, average velocities were higher within and proximal to the lesion compared to distal regions. Lesions with greater cross-sectional area exhibited less proximal recovery, with velocity remaining elevated rather than returning to baseline levels, an effect that was more pronounced at higher inflow rates. Color Doppler studies revealed a flow eddy at the lesion inlets, consistent with clinical observations in patients with sudden stenosis. Particle-tracking further supported these findings. These results suggest that the degree of occlusion (i.e., cross-sectional area) influences flow dynamics more than the length of the lesion. Additionally, this study demonstrates the feasibility of replicating clinically observed venous flow patterns within a benchtop setting. The ultrasound-compatible models developed provide greater anatomical relevance with the inclusion of multiple source veins, making them more accurate for acquiring flow measurements and valuable for future venous disease research.</p>

Des colloïdes à base de silice et de polystyrène ont été synthétisés. Les particules d’oxyde ont d’abord été élaborées et modifiées en surface, puis ont servi de germes au cours d’une étape de polymérisation du styrène. Deux procédés de polymérisation en phase hétérogène ont été utilisés (émulsion ou dispersion) menant à des colloïdes aux morphologies originales et contrôlées. Une étude morphologique par tomographie électronique a permis de mieux comprendre les mécanismes de croissance et d’organisation des particules de latex autour des germes de silice. La synthèse de particules Janus pour l’imagerie biomédicale est aussi décrite. Ces particules de silice ont été modifiées en surface par un chromophore biphotonique et un agent de reconnaissance de certaines cellules tumorales. Des études spectroscopiques et des tests de cytotoxicité ont été entrepris. / Hybrid colloids based on silica and polystyrene have been synthesized. Oxide particles were first elaborated, surface modified, and then used as seed in a styrene polymerization step. Two heterogeneous polymerisation proceeds were employed (emulsion or dispersion) leading to colloids with original and controlled morphologies. A morphological study by electronic tomography enabled to better understand growth and organisation mechanisms of latexes around silica seeds. Janus particles synthesis for biomedical imaging is also described. Silica particles were surface modified with a biphotonic chromophore and a tumor cells targeting agent. Spectroscopic studies and cytotoxicity tests were investigated.

La segmentation de régions d'intérêt dans le cadre de l'analyse d'images médicales et biomédicales reste encore à ce jour un challenge en raison notamment de la variété des modalités d'acquisition et des caractéristiques associées (bruit par exemple).Dans ce contexte particulier, cet exposé présente une méthode de segmentation de type contour actif dont l ‘énergie associée à l'obtention de l'équation d'évolution s'appuie sur une mesure de similarité entre les densités de probabilités (en niveau de gris) des régions intérieure et extérieure au contour au cours du processus itératif de segmentation. En particulier, nous nous intéressons à la famille particulière des alpha-divergences. L'intérêt principal de cette méthode réside (i) dans la flexibilité des alpha-divergences dont la métrique intrinsèque peut être paramétrisée via la valeur du paramètre alpha et donc adaptée aux distributions statistiques des régions de l'image à segmenter ; et (ii) dans la capacité unificatrice de cette mesure statistique vis-à-vis des distances classiquement utilisées dans ce contexte (Kullback- Leibler, Hellinger...). Nous abordons l'étude de cette mesure statistique tout d'abord d'un point de vue supervisé pour lequel le processus itératif de segmentation se déduit de la minimisation de l'alpha-divergence (au sens variationnel) entre la densité de probabilité courante et une référence définie a priori. Puis nous nous intéressons au point de vue non supervisé qui permet de s'affranchir de l'étape de définition des références par le biais d'une maximisation de distance entre les densités de probabilités intérieure et extérieure au contour. Par ailleurs, nous proposons une démarche d'optimisation de l'évolution du paramètre alpha conjointe au processus de minimisation ou de maximisation de la divergence permettant d'adapter itérativement la divergence à la statistique des données considérées. Au niveau expérimental, nous proposons une étude comparée des différentes approches de segmentation : en premier lieu, sur des images synthétiques bruitées et texturées, puis, sur des images naturelles. Enfin, nous focalisons notre étude sur différentes applications issues des domaines biomédicaux (microscopie confocale cellulaire) et médicaux (radiographie X, IRM) dans le contexte de l'aide au diagnotic. Dans chacun des cas, une discussion sur l'apport des alpha-divergences est proposée. / In the particular field of Computer-Aided-Diagnosis, the segmentation of particular regions of interest corresponding usually to organs is still a challenging issue mainly because of the various existing for which the charateristics of acquisition are very different (corrupting noise for instance). In this context, this PhD work introduces an original histogram-based active contour segmentation using alpha-divergence family as similarity measure. The method keypoint are twofold: (i) the flexibility of alpha-divergences whose metric could be parametrized using alpha value can be adaptedto the statistical distribution of the different regions of the image and (ii) the ability of alpha-divergence ability to enbed standard distances like the Kullback-Leibler's divergence or the Hellinger's one makes these divergences an interesting unifying tool.In this document, first, we propose a supervised version of proposed approach:. In this particular case, the iterative process of segmentation comes from alpha-divergenceminimization between the current probability density function and a reference one which can be manually defined for instance. In a second part, we focus on the non-supervised version of the method inorder to be able.In that particular case, the alpha-divergence maximization between probabilitydensity functions of inner and outer regions defined by the active contour is maximized. In addition, we propose an optimization scheme of the alpha parameter jointly with the optimization of the divergence in order to adapt iteratively the divergence to the inner statistics of processed data. Furthermore, a comparative study is proposed between the different segmentation schemes : first, on synthetic images then, on natural images. Finally, we focus on different kinds of biomedical images (cellular confocal microscopy) and medical ones (X-ray) for computer-aided diagnosis.