Inflammatory Bowel Disease (IBD) is a chronic inflammatory condition of the gastrointestinal tract, often associated with reduced food intake (anorexia) and weight loss. The anorexia manifest following gastrointestinal inflammation can only be expressed if appropriate signals are communicated from the inflamed segment to the brain. Yet, the nature of these signals, and the identity of the brain sites processing these anorexigenic signals, are unknown. The present experiment evaluates the contribution of the paraventricular nucleus (PVN), a brain site rich in corticotropin releasing factor (CRF) receptors and known to be involved in the control of food intake, in the anorexia associated with experimental colitis. Colitis was induced, by trinitrobenzenesulfonic acid (TNB) treatment, in animals in which the PVN was ablated or in rats with sham brain surgeries. Results indicated clearly that the expression of the anorexia following TNB treatment is fully expressed even in the absence of the PVN. This result indicates that the integrity of the PVN is not necessary for the reduction of eating associated with intestinal inflammation, thus suggesting that CRF is also not critical to colitis-induced anorexia. inflammatory bowel disease (IBD); feeding; anorexia; gut-brain communication; paraventricular nucleus (PVN); interleukin-1 (IL-l); corticotropin-releasing-factor (CRF); neuroimmunology / Thesis / Master of Science (MS)

Scanning probe microscopy (SPM) tip-based nanofabrication (TBN) is a technique that directly creates a variety of nanostructures on a substrate using the nanoscale probe tips. SPM TBN possesses superior resolution and flexibility: nanostructures with feature size under 5 nm have been achieved via SPM TBN, which is beyond what the state-of-the art optical-based lithography technique can provide. However, the inherent serial nature of SPM TBN makes it a low throughput process. Multi-probe SPM systems have therefore been developed to increase the nanofabrication efficiency. Atomic force microscopy (AFM) and scanning tunneling microscopy (STM) are two most commonly used SPM TBN techniques. Most of prior work has focused on contact-mode AFM-based TBN. This work, using CMOS MEMS technology as the design and fabrication platform, develops an active conductive probe array that aims to perform parallel surface imaging and nanofabrication in non-contact STM mode. The CMOS-MEMS process provides a monolithic integration of MEMS devices with CMOS electronics that can facilitate future automation and parallel probe operation. The CMOS-MEMS probe adopts a micro-cantilever structure and applies bimorph electrothermal actuation to control the vertical displacement of the probe tips. The cantilever is designed to be stiff, with a spring constant of 36 N/m that is larger than the force gradient of the cantilever tip-sample interaction forces in the working distance regime of STM in order to avoid the tip-to-sample “snap-in” and ensure the stability of the STM feedback system. A modified Spindt tip process, compatible with post-CMOS MEMS processing, is developed to batch fabricate Ni/Pt composite tips on CMOS-MEMS probe arrays that are used as STM end-effectors. The integrated Ni/Pt tips on the MEMS probes have a tip radius down to 50 vii nm. The Spindt tip demonstrates the capability of both imaging and nanowire fabrication in STM mode. A hierarchical dual-servo STM system is constructed for the parallel STM imaging using two CMOS-MEMS probes. The system consists of a piezoelectric actuator-driven servo and an electrothermal actuator-driven servo to control the vertical displacement of two probe tips and maintain a constant current between the tips and the sample. Both servos use a proportionalintegral controller. The dual-servo STM system is capable of parallel STM image acquisition using CMOS MEMS probe arrays. An on-chip electrothermal proximity sensor pair and probes with embedded microgoniometers are designed to assist the alignment between the CMOS-MEMS probe array and the examined sample surface. The electrothermal proximity sensor pair is used to measure the separation and the non-parallelism between the probe chip and the sample. The electrothermal proximity sensor has a positioning accuracy of around 1 μm. An electrothermal microgoniometer platform is developed to hold a one-dimensional array of active CMOS-MEMS probes and serves to provide the in situ fine adjustment of relative height among these probes. The micro-goniometer has a maximum tilt of 1.2°, which is sufficient to compensate the probe chip-sample misalignment and the possible height difference among array probes introduced by process variations.

Les réseaux bayésiens dynamiques (RBD) sont une classe de modèles graphiques probabilistes qui est devenu un outil standard pour la modélisation de divers phénomènes stochastiques variant dans le temps. A cause de la complexité induite par l'ajout de la dimension temporelle, l'apprentissage de la structure DBN est une tâche très complexe. Les algorithmes existants sont des adaptations des algorithmes d'apprentissage de structure pour les RB basés sur score mais sont souvent limités lorsque le nombre de variables est élevé. Une autre limitation pour les études d'apprentissage de la structure des RBD, ils utilisent leurs propres Benchmarks et techniques pour l' évaluation. Le probl ème dans le cas dynamique, nous ne trouvons pas de travaux antérieurs qui fournissent des détails sur les réseaux et les indicateurs de comparaison utilisés. Nous nous concentrons dans ce projet à l'apprentissage de la structure des RBD et ses méthodes d'évaluation avec respectivement une autre famille des algorithmes d'apprentissage de la structure, les méthodes de recherche locale, et une nouvelle approche de génération des grandes standard RBD et une métrique d'évaluation. Nous illustrons l'intérêt de ces méthodes avec des résultats expérimentaux.

No description available.