Functional magnetic resonance imaging (fMRI) can detect neuronal activations in the entire brain, in response to a stimulus, through the observation of subsequent hemodynamic changes. In particular, fMRI is a good tool for studying the mechanisms of deep brain stimulation and vagus nerve stimulation, which are still poorly understood. However, this technique is not readily usable in humans because of safety concerns towards the action of the strong magnetic fields used in MRI on implanted electrodes. Indeed, methodological developments in animals are needed. The main goal of this thesis is to study the mechanisms of central and peripheral nervous system stimulation in rats by fMRI. First, fast MRI sequences used in fMRI are exposed, such as multishot Echo-Planar Imaging, allowing to image the entire brain in a couple of seconds. Various imaging problems posed by these sequences, such as magnetic susceptibility artifacts, are also presented. These sequences, developed during this thesis, associated with electrophysiological measurements, allowed imaging of epileptic networks in the rat. Secondly, animal preparation is developped, as it is peculiar in fMRI : neuronal activations, as well as neurovascular coupling, must be preserved under anesthesia. Compared to anesthesia by isoflurane and ketamine, it was concluded that medetomidine was an anesthetic of choice for fMRI of the rodent, and the protocol used for animal preparation for imaging is specified. Furthermore, the electrodes used in deep brain stimulation induce significant artifacts in MRI images, and electrodes made of amagnetic materials are needed. Our choice of carbon electrodes is explained, and the manufacturing protocol used is exposed. These methodological developments were then validated in fMRI experiments of hypercapnic challenges and forepaw stimulation. Finally, an fMRI experiment studying mechanisms of action of vagus nerve stimulation was conducted, focusing on the distinction between neuronal activations and confounding cardiovascular effects by dynamic causal modeling. Also, results on fMRI of deep brain stimulation in rats are presented. Several targets were stimulated (dorsolateral geniculate nucleus, dentate gyrus, striatum and thalamus), and activations were obtained at a distance from the electrode. Results were in accordance with current knowledge on neuroanatomical connections of these nuclei. Thus, we developed and validated fMRI of the rat and its application to electrical stimulation of peripheral and central nervous system.
