Magnetic resonance spectroscopy (MRS) is a powerful tool when studying metabolism in intact and living organs preserving cells in their natural microenvironment with minimal external interference. The unique insights offered by the observation of metabolic processes in vivo are available thanks to the non-invasiveness of the MR technique, due to the employment of non-ionizing radiations at room temperature. These conditions are ideal when studying cerebral metabolism in vivo, since the brain is a particularly complex organ where neurons and glial cells are highly interconnected and dependent from each other. Moreover, brain cells are protected by the blood-brain barrier, which regulates the delivery of nutrients and chemicals. The work of this thesis focused on the application of 1H and 13C MRS in vivo to study cerebral metabolism in mice. 18F-Fluorodeoxyglucose (18F-FDG) positron emission tomography (PET) was used in complement to the abovementioned MR techniques to determine the cerebral metabolic rate of glucose (CMRglc) or local glucose uptake under specific physiological and pathological conditions. At first, we validated the application of 13C MRS to small volumes in the mouse brain in combination to compartmental modeling for neuro-glial metabolism, as previously applied to humans and rats in larger volumes. The second part of the thesis focused on the translation of these techniques to the study of brain metabolism during glioblastoma invasion. A highly infiltrative mouse xenograft model of glioblastoma derived from human glioma spheres was employed in our study to determine the evolution over time of the neurochemical profile detected with 1H MRS at 14.1 Tesla. 13C-labeling experiments with [1,6-13C]glucose were performed at late stage after extensive glioma invasion in the whole brain. Metabolic fluxes were determined under these conditions reflecting energetic metabolism in neurons and glioma/glial cells. Globally, infiltrating cells did not show a highly glycolytic metabolism, as often observed in cancer cells and in glioma cells close to the tumor core, suggesting that environmental stimuli induced by a central necrosis, high cellular density and poor vascularization may be crucial to the occurrence of a metabolic switch. Finally, a novel patient-derived mouse xenograft model of glioblastoma was tested and characterized longitudinally with 1H MRS at 14.1 Tesla. Glioma cells were implanted intracranially after patient tumor resection and minimal manipulation in vitro in order to minimize a selective pressure and promote a close resemblance of the xenograft with the parental tumor. This study showed that patient-derived xenografts obtained from freshly injected glioma cells offer a reliable model to study glioma invasion and related treatments while raising intriguing questions on the mechanisms of adaptation of human cells to the brain of the host.