No description available.

Le dioxyde de carbone (CO2) est un résidu naturel du métabolisme cellulaire, la troisième substance la plus abondante du sang, et un important agent vasoactif. À la moindre variation de la teneur en CO2 du sang, la résistance du système vasculaire cérébral et la perfusion tissulaire cérébrale subissent des changements globaux. Bien que les mécanismes exacts qui sous-tendent cet effet restent à être élucidés, le phénomène a été largement exploité dans les études de réactivité vasculaire cérébrale (RVC). Une voie prometteuse pour l’évaluation de la fonction vasculaire cérébrale est la cartographie de la RVC de manière non-invasive grâce à l’utilisation de l’Imagerie par Résonance Magnétique fonctionnelle (IRMf). Des mesures quantitatives et non-invasives de de la RVC peuvent être obtenus avec l’utilisation de différentes techniques telles que la manipu- lation du contenu artériel en CO2 (PaCO2) combinée à la technique de marquage de spin artériel (Arterial Spin Labeling, ASL), qui permet de mesurer les changements de la perfusion cérébrale provoqués par les stimuli vasculaires. Toutefois, les préoccupations liées à la sensibilité et la fiabilité des mesures de la RVC limitent de nos jours l’adoption plus large de ces méthodes modernes de IRMf. J’ai considéré qu’une analyse approfondie ainsi que l’amélioration des méthodes disponibles pourraient apporter une contribution précieuse dans le domaine du génie biomédical, de même qu’aider à faire progresser le développement de nouveaux outils d’imagerie de diagnostique. Dans cette thèse je présente une série d’études où j’examine l’impact des méthodes alternatives de stimulation/imagerie vasculaire sur les mesures de la RVC et les moyens d’améliorer la sensibilité et la fiabilité de telles méthodes. J’ai aussi inclus dans cette thèse un manuscrit théorique où j’examine la possible contribution d’un facteur méconnu dans le phénomène de la RVC : les variations de la pression osmotique du sang induites par les produits de la dissolution du CO2. Outre l’introduction générale (Chapitre 1) et les conclusions (Chapitre 6), cette thèse comporte 4 autres chapitres, au long des quels cinq différentes études sont présentées sous forme d’articles scientifiques qui ont été acceptés à des fins de publication dans différentes revues scientifiques. Chaque chapitre débute par sa propre introduction, qui consiste en une description plus détaillée du contexte motivant le(s) manuscrit(s) associé(s) et un bref résumé des résultats transmis. Un compte rendu détaillé des méthodes et des résultats peut être trouvé dans le(s) dit(s) manuscrit(s). Dans l’étude qui compose le Chapitre 2, je compare la sensibilité des deux techniques ASL de pointe et je démontre que la dernière implémentation de l’ASL continue, la pCASL, offre des mesures plus robustes de la RVC en comparaison à d’autres méthodes pulsés plus âgées. Dans le Chapitre 3, je compare les mesures de la RVC obtenues par pCASL avec l’utilisation de quatre méthodes respiratoires différentes pour manipuler le CO2 artérielle (PaCO2) et je démontre que les résultats peuvent varier de manière significative lorsque les manipulations ne sont pas conçues pour fonctionner dans l’intervalle linéaire de la courbe dose-réponse du CO2. Le Chapitre 4 comprend deux études complémentaires visant à déterminer le niveau de reproductibilité qui peut être obtenu en utilisant des méthodes plus récentes pour la mesure de la RVC. La première étude a abouti à la mise au point technique d’un appareil qui permet des manipulations respiratoires du CO2 de manière simple, sécuritaire et robuste. La méthode respiratoire améliorée a été utilisée dans la seconde étude – de neuro-imagerie – où la sensibilité et la reproductibilité de la RVC, mesurée par pCASL, ont été examinées. La technique d’imagerie pCASL a pu détecter des réponses de perfusion induites par la variation du CO2 dans environ 90% du cortex cérébral humain et la reproductibilité de ces mesures était comparable à celle d’autres mesures hémodynamiques déjà adoptées dans la pratique clinique. Enfin, dans le Chapitre 5, je présente un modèle mathématique qui décrit la RVC en termes de changements du PaCO2 liés à l’osmolarité du sang. Les réponses prédites par ce modèle correspondent étroitement aux changements hémodynamiques mesurés avec pCASL ; suggérant une contribution supplémentaire à la réactivité du système vasculaire cérébral en lien avec le CO2. / Carbon dioxide (CO2) is a natural byproduct of cellular metabolism, the third most abundant substance of blood, and a potent vasoactive agent. The resistance of cerebral vasculature and perfusion of the brain tissue respond to the slightest change in blood CO2 content. The physiology of such an effect remains elusive, yet the phenomenon has been widely exploited in studies of the cerebral vascular function. A promising avenue for the assessment of brain’s vascular function is to measure the cerebrovascular reactivity to CO2 (CVR) non-invasively using functional MRI. Quantitative and non-invasive mapping of CVR can be obtained using respiratory manipulations in arterial CO2 and Arterial Spin Labeling (ASL) to measure the perfusion changes associated with the vascular stimulus. However, concerns related to the sensitivity and reliability of CVR mea- sures by ASL still limit their broader adoption. I considered that a thorough analysis and amelioration of available methods could bring a valuable contribution in the domain of biomedical engineering, helping to advance new diagnostic imaging tools. In this thesis I present a series of studies where I exam the impact of alternative manipulation/ASL methods on CVR measures, and ways to improve the sensitivity and reliability of these measures. I have also included in this thesis a theoretical paper, where I exam the possible contribution of an unappreciated factor in the CVR phenomenon: the changes in blood osmotic pressure induced by the products of CO2 dissolution. Apart from a general introduction (Chapter 1) and conclusion (Chapter 6), this thesis comprises 4 other chapters, in which five different research studies are presented in the form of articles accepted for publication in scientific journals. Each of these chapters begins with its own specific introduction, which consists of a description of the background motivating the study and a brief summary of conveyed findings. A detailed account of methods and results can be found in the accompanying manuscript(s). The study composing Chapter 2 compares the sensitivity of two state-of-the-art ASL techniques and show that a recent implementation of continuous ASL, pCASL, affords more robust measures of CVR than older pulsed methods. The study described in Chapter 3 compares pCASL CVR measures obtained using 4 different respiratory methods to manipulate arterial CO2 (PaCO2) and shows that results can differ significantly when manipulations are not designed to operate at the linear range of the CO2 dose-response curve. Chapter 4 encompasses two complementary studies seeking to determine the degree of reproducibility that can be attained measuring CVR using the most recent methods. The first study resulted in the technical development of a breathing apparatus allowing simple, safe and robust respiratory CO2 manipulations. The improved respiratory method was used in the second – neuroimaging – study, in which I and co-authors investigate the sensitivity and reproducibility of pCASL measuring CVR. The pCASL imaging technique was able to detect CO2-induced perfusion responses in about 90% of the human brain cortex and the reproducibility of its measures was comparable to other hemodynamic measures already adopted in the clinical practice. Finally, in Chapter 5 I present a mathematical model that describes CVR in terms of PaCO2-related changes in blood osmolarity. The responses predicted by this model correspond closely to the hemodynamic changes measured with pCASL, suggesting an additional contribution to the reactivity of cerebral vasculature to CO2.

Le flux sanguin cérébral (CBF) caractérise la micro-circulation et l'irrigation des tissus. Cette information de perfusion cérébrale est utilisé en clinique pour le diagnostic et le suivi thérapeutique de nombreuses maladies. La technique de mesure de CBF la moins invasive est celle par marquage de spins artériels (ASL) où l'eau du sang fait office de traceur. L'objectif de cette thèse, menée dans le cadre d'une convention CIFRE, consistait à faciliter l'utilisation de séquences ASL continues et pseudo-continues (CASL, pCASL) ainsi qu'à améliorer leur performance en pré-clinique. En effet, la mesure quantitative de CBF par ASL est un protocole complexe qui nécessite plusieurs étapes d'ajustements, d'acquisitions et de traitement de données. Dans le but d'alléger ce protocole, un package CASL a été développé en collaboration avec Bruker. Plusieurs étapes d'ajustements et de post-processing ont été automatisées, rendant la génération de cartes CBF relatives et absolues plus aisée. Le champ magnétique élevé des scanners IRM pré-cliniques présente de nombreux avantages mais est également une source de problèmes en ASL. Nous nous sommes intéressés plus particulièrement à deux d'entre eux : l'instabilité du marquage de spins et l'échauffement induit par les séquences ASL. Pour stabiliser le marquage ASL, une stratégie d'optimisation de la séquence pCASL a été développée et testée chez le rat à 9.4 T. Ceci a permis l'obtention d'un marquage robuste, même en situations de shim dégradé. Le package pCASL a été partagé avec dix autres instituts dans le monde. L'échauffement induit lors de séquences CASL et pCASL par le dépôt d'énergie radiofréquence a été caractérisé globalement et localement, dans le cerveau et au niveau des carotides, pour deux configurations d'antenne d'émission. Pour finir, une séquence pCASL encodée en temps a été développée et appliquée à la souris, dans le cadre d'une collaboration avec des équipes néerlandaises du Leiden University Medical Center. Cet outil permet la mesure simultanée de CBF et du temps de transit artériel, un paramètre pouvant refléter des pathologies vasculaires sous-jacentes. / Cerebral blood flow (CBF) characterizes the blood supply to brain tissue. This perfusion-related parameter contributes in diagnosis and therapeutic follow-up in many diseases. The least invasive technique to measure CBF is arterial spin labeling (ASL), where arterial water is used as tracer. The aim of this PhD project, conducted within a CIFRE agreement (Convention Industrielle de Formation par la REcherche), was to increase the performance and to facilitate the use of continuous and pseudo-continuous arterial spin labeling (CASL, pCASL) tools in preclinical studies. CBF quantification by means of ASL is one of the most challenging MRI modalities in terms of the workflow, since additional adjustments, acquisitions and post-processing steps are required. First, to render the workflow smoother for the user, a CASL package has been developed in collaboration with Bruker. This workflow allows easier relative and absolute CBF measurements, thanks to the integration of automated adjustments and reconstruction steps. In a second step, problems arising at high magnetic field were addressed. A strategy to optimize the pCASL labeling sequence in order to obtain robust results was developed and its robustness towards suboptimal shim conditions was demonstrated at 9.4 T in rats. The developed pCASL-package, consisting of three sequences, was shared with ten other institutes worldwide. Another issue encountered at high magnetic fields is heating due to RF power deposition, which was assessed locally in the brain and in the carotids, as well as globally, for the CASL and pCASL sequences and for two different transmit coil configurations. In a third step, time-encoded pCASL was developed in mice in collaboration with teams of the Leiden University Medical Center. This tool enables the simultaneous mapping of CBF and arterial transit time, a parameter that can reflect underlying pathologies such as increased vessel tortuosity or occlusion.

MRI can provide anatomical, functional, and physiological images at relatively high spatial resolution and is non-invasive and does not have depth limitation. However, the application of MRI to study the retina is difficult due to the very small size of the retina. This thesis details the development of MRI methods to image blood flow (BF), anatomy, and function of the retina and choroid, and their application to two diseases of the retina: diabetic retinopathy and retinal degeneration. A unique continuous arterial spin labeling technique was developed to image BF in mice and tested by imaging cerebral BF. This method was then applied to image layer-specific BF of the retina and choroid in mice, and to acquire BF functional MRI of the retina and choroid in response to hypoxic challenge. Additionally blood oxygen level dependent functional MRI of the mouse retina and choroid in response to hypoxic challenge was obtained using a balanced steady state free precession sequence which provides fast acquisition, has high signal to noise ratio, and does not have geometric distortion or signal dropout artifacts. In a mouse model of diabetic retinopathy, MRI detected reduced retinal BF in diabetic animals. Visual function in the diabetic mice, as determined by psychophysical tests, was also reduced. Finally, in a mouse model of retinal degeneration, BF and anatomical MRI detected reductions of retinal BF and the thickness of the retina. The studies detailed in this thesis demonstrate the feasibility of layer-specific MRI to study BF, anatomy, and function, in the mouse retina. Further, these methods were shown to provide a novel means of studying animal models of retinal disease in vivo.

Microvascular lesions of the brain are observed in numerous pathological conditions including Alzheimer's disease (AD). Regional patterns of microvascular abnormality can be characterized using current neuroimaging technologies. When applied to mouse models of human disease, these technologies reveal cerebral vascular patterns and help uncover genotype-to-phenotype relationships. This thesis focuses on the development and testing of techniques for measuring two perfusion-related metrics in mouse brain regions, namely, cerebral blood volume (CBV) and cerebral blood flow (CBF) using micro-computed tomography (micro-CT) and arterial spin labeling (ASL), respectively. The main developments for measurement of CBV have included: refinements to micro-CT specimen preparation; registration of micro-CT images to an MRI anatomical brain atlas; and masking of major vessels to calculate small-vessel CBV (sv-CBV). The development of this micro-CT technique provided reference values of CBV over neuroanatomical brain regions in wildtype mice. A separate study was conducted to assess regional sv-CBV in a mouse model of AD; this study was motivated by the prevalence of microvascular lesions in patients who suffer from AD. Significant regional differences in sv-CBV were found between AD-afflicted mice and controls. The main developments for measurement of CBF have included: design and implementation of accurate ASL slice positioning and optimization of inversion efficiency parameters. The development of this ASL technique provided reference values of CBF over neuroanatomical brain regions in wildtype mice. These techniques for measuring CBV and CBF over mouse brain regions could lead to improved characterization of vascularity in models of neurological diseases.

Microvascular lesions of the brain are observed in numerous pathological conditions including Alzheimer's disease (AD). Regional patterns of microvascular abnormality can be characterized using current neuroimaging technologies. When applied to mouse models of human disease, these technologies reveal cerebral vascular patterns and help uncover genotype-to-phenotype relationships. This thesis focuses on the development and testing of techniques for measuring two perfusion-related metrics in mouse brain regions, namely, cerebral blood volume (CBV) and cerebral blood flow (CBF) using micro-computed tomography (micro-CT) and arterial spin labeling (ASL), respectively. The main developments for measurement of CBV have included: refinements to micro-CT specimen preparation; registration of micro-CT images to an MRI anatomical brain atlas; and masking of major vessels to calculate small-vessel CBV (sv-CBV). The development of this micro-CT technique provided reference values of CBV over neuroanatomical brain regions in wildtype mice. A separate study was conducted to assess regional sv-CBV in a mouse model of AD; this study was motivated by the prevalence of microvascular lesions in patients who suffer from AD. Significant regional differences in sv-CBV were found between AD-afflicted mice and controls. The main developments for measurement of CBF have included: design and implementation of accurate ASL slice positioning and optimization of inversion efficiency parameters. The development of this ASL technique provided reference values of CBF over neuroanatomical brain regions in wildtype mice. These techniques for measuring CBV and CBF over mouse brain regions could lead to improved characterization of vascularity in models of neurological diseases.

Arterial spin labeling (ASL) perfusion Magnetic Resonance Imaging (MRI) is a noninvasive technique for measuring quantitative cerebral blood flow (CBF) but subject to an inherently low signal-to-noise-ratio (SNR), resulting in a big challenge for data processing. Traditional post-processing methods have been proposed to reduce artifacts, suppress non-local noise, and remove outliers. However, these methods are based on either implicit or explicit models of the data, which may not be accurate and may change across subjects. Deep learning (DL) is an emerging machine learning technique that can learn a transform function from acquired data without using any explicit hypothesis about that function. Such flexibility may be particularly beneficial for ASL denoising. In this dissertation, three different machine learning-based methods are proposed to improve the image quality of ASL MRI: 1) a learning-from-noise method, which does not require noise-free references for DL training, was proposed for DL-based ASL denoising and BOLD-to-ASL prediction; 2) a novel deep learning neural network that combines dilated convolution and wide activation residual blocks was proposed to improve the image quality of ASL CBF while reducing ASL acquisition time; 3) a prior-guided and slice-wise adaptive outlier cleaning algorithm was developed for ASL MRI. In the first part of this dissertation, a learning-from-noise method is proposed for DL-based method for ASL denoising. The proposed learning-from-noise method shows that DL-based ASL denoising models can be trained using only noisy image pairs, without any deliberate post-processing for obtaining the quasi-noise-free reference during the training process. This learning-from-noise method can also be applied to DL-based ASL perfusion prediction from BOLD fMRI as ASL references are extremely noisy in this BOLD-to-ASL prediction. Experimental results demonstrate that this learning-from-noise method can reliably denoise ASL MRI and predict ASL perfusion from BOLD fMRI, result in improved signal-to-noise-ration (SNR) of ASL MRI. Moreover, by using this method, more training data can be generated, as it requires fewer samples to generate quasi-noise-free references, which is particularly useful when ASL CBF data are limited. In the second part of this dissertation, we propose a novel deep learning neural network, i.e., Dilated Wide Activation Network (DWAN), that is optimized for ASL denoising. Our method presents two novelties: first, we incorporated the wide activation residual blocks with a dilated convolution neural network to achieve improved denoising performance in term of several quantitative and qualitative measurements; second, we evaluated our proposed model given different inputs and references to show that our denoising model can be generalized to input with different levels of SNR and yields images with better quality than other methods. In the final part of this dissertation, a prior-guided and slice-wise adaptive outlier cleaning (PAOCSL) method is proposed to improve the original Adaptive Outlier Cleaning (AOC) method. Prior information guided reference CBF maps are used to avoid bias from extreme outliers in the early iterations of outlier cleaning, ensuring correct identification of the true outliers. Slice-wise outlier rejection is adapted to reserve slices with CBF values in the reasonable range even they are within the outlier volumes. Experimental results show that the proposed outlier cleaning method improves both CBF quantification quality and CBF measurement stability. / Electrical and Computer Engineering

A novel non-invasive imaging method of unique k-space trajectory named “3D center-out EPI with cylindrical encoding” was developed and implemented for fast imaging of the human brain. The method based on a variant of 3D hybrid EPI combines advantages of the Cartesian and the radial encoding to achieve ultra-short echo time independent of spatial resolution and reasonably short echo train length yielding a quality image of high signal-to-noise ratio. Unlike rectilinear sampling, the method offers not only less motion and flow artifacts but enables also the undersampling capability. As a result, the method improves temporal resolution by shortening the measurement time. Nonetheless, artifacts induced from long-term drifts of the magnetic field as well as geometrical distortions caused by B0 inhomogeneity were removed with the average phase of the k-space center lines and an additional field map scan. Compared to other cylindrical k-space trajectories based on echo-planar imaging, which lead to progressively increasing echo time upon increasing the spatial resolution, the proposed method offers more benefits. As a significant application, imaging readout of the novel technique was applied to true 3D cine imaging which was later used in the combination of pseudo-continuous arterial spin labeling module in order to track a short arterial spin labeling (ASL) bolus of well-defined length along the fast passage through the large vessel compartment of the brain. Parametric maps of ASL signal change, estimated time-to-peak and ASL bolus width were extracted in order to characterize the macrovascular compartments of the brain-feeding arteries. Consequently, bolus dispersion within a single arterial branch was also assessed.

La perfusion est un facteur important dans la viabilité et la fonction du myocarde. Des atteintes microvasculaires diffuses, précédant l'infarctus ou l'insuffisance cardiaque sont impliqués dans bon nombre de pathologies cardiaques. Ce travail vise à améliorer les techniques existantes de mesure quantitatives et non-invasive de la perfusion myocardique par marquage de spins artériels (ASL). La première partie de mon travail de thèse a consisté en la mise place chez la souris d'une technique alternative pour mesurer la perfusion myocardique. Celle-ci est basée sur un marquage pulsé et régulièrement répété afin de construire un état d'équilibre de l'aimantation sous l'influence de la perfusion (approche steady-pulsed ASL). Le modèle théorique associé à cette technique spASL a été développé en parallèle afin de quantifier le flux sanguin tissulaire. Il a été montré que spASL permettait d'obtenir un résultat similaire aux techniques existantes avec en plus, les avantages d'améliorer la sensibilité au signal de perfusion ainsi que de réduire le temps d'acquisition. Dans un second temps, un transfert vers l'imagerie clinique pour une application chez l'homme a été entrepris. Le marquage de type spASL a été conservé et le module de lecture a été adapté aux spécificités de l'imagerie cardiaque chez l'homme pour une acquisition en respiration libre. Un post-traitement dédié qui comprend une correction de mouvement rétrospective a ensuite vu le jour afin d'améliorer la robustesse de nos mesures. Parallèlement aux développements conduits chez l'homme, nous avons exploité l'approche spASL chez l'animal en proposant diverses améliorations en fonction des études menées. / Myocardial blood flow is an important factor of tissue viability and function. Diffuse changes in microcirculation preceding heart failure are involved in various cardiac pathologies. This work aim at improving existing techniques allowing quantitative and non-invasive myocardial perfusion assessment using arterial spin labeling. The first step of my work was to design an alternative approach to quantify myocardial blood flow in mice. The so called steady-pulsed ASL (spASL) is based on a regularly repeated pulsed labeling in order to build up a stationary regime of the magnetization under the influence of perfusion. The associated theoretical model has been developed in parallel to quantify tissue blood flow. We have shown that spASL allows to obtain similar results than the previously employed techniques, with the additional advantages of an increased sensitivity to the perfusion signal and a reduced acquisition time. A transfer towards clinical imaging for human applications was then undertaken. The spASL labeling scheme has been preserved while adapting the readout module to the specificities of cardiac MRI when applied to free-breathing human acquisitions. A dedicated post-processing, which includes a retrospective motion correction, has emerged subsequently to improve the robustness of our measurements. In parallel to the developments made for human studies, some optimization of the spASL technique when applied to rodent have been carried out depending on the conducted studies.

A esquizofrenia é um transtorno psiquiátrico incapacitante que afeta estimadamente 1% da população mundial. Delírios, alucinações, desorganização de pensamento e prejuízo cognitivo são as principais marcas da Esquizofrenia. Fisiologicamente, além de anormalidades funcionais e estruturais, alterações na atividade neuronal são reportadas. Como a atividade neuronal possui uma relação direta com o fluxo sanguíneo cerebral (CBF, Cerebral Blood Flow), a técnica de Imagens por Ressonância Magnética, denominada Marcação dos Spins Arteriais (ASL, Arterial Spin Labeling), que permite a obtenção de mapa quantitativo de CBF, é uma ferramenta útil na avaliação funcional cerebral. Além disso, a ASL pode ser usada na avaliação da conectividade funcional, que é eficiente na investigação de rupturas funcionais entre as regiões do cérebro. Comparando com um grupo de sujeitos saudáveis, os pacientes com esquizofrenia, recrutados no Hospital das Clínicas de Ribeirão Preto (HCFMRP), apresentaram redução de CBF em regiões bilaterais do polo frontal e giro frontal superior, giro frontal medial direito, partes triangular e opercular do giro frontal inferior direito, divisão posterior do giro supramarginal esquerdo, divisão superior e inferior do córtex occipital lateral esquerdo e polo occipital. A conectividade funcional, avaliada por três diferentes métodos (baseado em semente, análise de componentes independentes e teoria dos grafos), se apresentou prejudicada em regiões envolvendo funções motoras, sensoriais e cognitivas dos pacientes. Portanto, utilizando uma técnica de imagem completamente não invasiva, foi possível observar déficits de CBF e alterações na organização funcional do cérebro de pacientes com esquizofrenia, relacionados com os sintomas e características da psicopatologia. / Schizophrenia is a disabling psychiatric disorder that affects around 1% of the population worldwide. Delusions, hallucinations, disorganized thought, and cognitive deficits are the main features of schizophrenia. Physiologically, in addition to functional and structural abnormalities, changes in neuronal activity are reported. Since the Cerebral Blood Flow (CBF) is directly related with neuronal activity, the Magnetic Resonance Imaging (MRI) technique called Arterial Spin Labeling (ASL), which allows the quantification of CBF, is a useful tool in brain functional evaluation. In addition, ASL can be used to assess functional connectivity, which is efficient in investigating functional impairment between regions of the brain. Patients with Schizophrenia, recruited at the Clinical Hospital (HCFMRP), presented a reduction of CBF in bilateral regions of the frontal pole and superior frontal gyrus, right medial frontal gyrus, triangular and opercular parts of the right inferior frontal gyrus, posterior division of left supramarginal gyrus, superior and inferior division of left lateral occipital cortex and occipital pole. Functional connectivity, assessed by three different methods (seed-based, independent component analysis and graph theory), was impaired in regions involving patients\' motor, sensory and cognitive functions. Therefore, using a noninvasive imaging technique, it was possible to observe CBF deficits and alterations in the functional organization of the brain of schizophrenia patients, related to the symptoms and characteristics of the psychopathology.