Development of enhanced and accelerated 2D and 3D 1H-MRSI at ultra-high field for mapping rodent brain metabolism

Magnetic Resonance Spectroscopic Imaging (MRSI) enables the simultaneous non-invasive acquisition of MR spectra from multiple spatial locations. This powerful tool has seen increase usage in different pathologies studies, among these are neurodegenerative diseases Yet, this spectroscopic technique is not yet widely applied in the preclinical setting, mostly because of difficulties specifically related to very small nominal voxel size that one can find in rodent studies and low concentration of brain metabolites, resulting in low signal-to-noise ratio (SNR). Additionally, the constant appetite for higher resolutions leads to an increase of acquisition time when using 1H-MRSI. In this context, the overall goal of the current thesis was to develop, implement and validate innovative fast 1H-MRSI techniques in the rat brain, tailor-made to enhance the SNR for ultra-high spatial resolution. For that we combined ultra-high field (UHF) scanners for increased SNR and spectral resolution, high performance radio-frequency coils, novel signal denoising approach, advanced and accelerated pulse sequences and processing/reconstruction tools developed in-house, thus going beyond the state-of-the-art in preclinical studies. This thesis explored the implementation and application of a free induction decay (FID) 1HMRSI sequence in the rat brain at 14.1 T. A repeatable acquisition protocol was presented in order to take advantages of the higher SNR and spectral resolution that comes with UHF MRSI. Furthermore, to facilitate the processing and the analysis, a user-friendly toolbox the MRS4Brain Toolbox-was programmed. Data acquired with the 1H-FID-MRSI showed high reproducibility with preservation of the brain metabolite distribution. Two denoising techniques, the Marchenko-Pastur principal component analysis (MP-PCA) and the Low-Rank Total Generalized Variations, were tested in order to evaluate their potential on preclinical 14.1 T fast in vivo 1H-FID-MRSI. As there is no ground truths for in vivo maps, Monte-Carlo simulations were created to observe the theoretical impacts of these two noise-reduction algorithms. While no alterations were observed within in vivo sets, the synthetic data showed that one should be mindful with the estimations for lower concentrated metabolites. Acceleration of the 1H-FID-MRSI at ultra-high field via compressed sensing (CS) was explored within 14.1 and 9.4T. The goal is to capitalize on the reduction of acquisition time to unlock the potential for higher resolution such as the 3D FID-MRSI methodology. For this purpose, the acceleration protocols were evaluated by estimating the impact of CS on the spatial resolution via the PSF and the appearance of artifacts on the spectra, relative to the gains in temporal resolution. The application of the optimized protocol allowed for the acquisition 3D and multi-slice 1H-FID-MRSI, with relative good quality spectra. Another acceleration tool was investigated with the Concentric Rings Trajectory