This thesis is composed of four studies centered on investigating cerebral metabolism using magnetic resonance spectroscopy (MRS) of hyperpolarized and non-hyperpolarized compounds at ultra-high field. In the first two chapters, we studied longitudinally the effects of different treatments for hepatic encephalopathy (HE) on the neurometabolic changes associated with chronic HE using a rat model of type C HE and a multimodal approach including 1H MRS. The effects of the antibiotic rifaximin administered at different doses as well as its combination with probiotics were studied in the established model of bile duct ligated (BDL) rats. We showed for the first time the beneficial effects of the combined rifaximin and the probiotic Vivomixx® on the neurometabolic changes in vivo and longitudinally in a rat model of type C HE. The longitudinal changes of some brain metabolites concentration (glutamine, glutamate, creatine), as well as gut bifidobacteria concentration, were significantly less pronounced in the group of rats treated with rifaximin and probiotics compared to non-treated rats. We also showed that rifaximin treatment alone had limited efficacy. When administered at human dose in rats, its effects appeared only at the early stages of the disease, whereas at a higher dose some neurometabolic changes associated with HE were attenuated but the general condition of the rats was worse. In the third study, we used dissolution dynamic nuclear polarization (DNP) to hyperpolarize [2H7, U-13C6]-D-glucose, which enabled us to monitor real-time glycolysis in the healthy mouse brain. Given that glucose metabolism is tightly linked to neuronal activity, we first investigated how the change in anesthesia impacts the cerebral metabolism of hyperpolarized glucose. Anesthetics are known to influence brain activity, and in this study we showed for the first time that switching from isoflurane anesthesia to a combination of lower doses of isoflurane and medetomidine had a high impact on cerebral glucose uptake and glycolytic flow, as reflected by the increased labelling of downstream [1-13C] lactate. The second part of this study aimed at evaluating the feasibility of quantifying cerebral glucose metabolism kinetics by evaluating two different mathematical models, where kinetic rate constants were determined by fitting the models on 13C curves by non-linear regression using the Levenberg-Marquardt algorithm. We showed that a 3-compartment model is more stable and reliable than a 4-compartment model. Finally, since lithium salts are widely used for treating bipolar disorder but its mechanism of action is still not clearly understood, we evaluated the potential of hyperpolarized 6Li injected at pharmacological doses to assess its real-time bio-distribution and pharmacokinetic in the rat brain. We demonstrated first that hyperpolarized 6Li can be detected at pharmacological concentration in the rat head with high signal-to-noise ratio (SNR). We also showed that the transport of lithium through the intact blood-brain barrier brain is a limiting factor and demonstrated a striking difference between apparent 6Li T1 values in a brain with normal or disrupted blood-brain barrier.