A New J-Resolved MRSI Framework for Whole-Brain Simultaneous Metabolite and Neurotransmitter Mapping

点击翻译按钮获取中文摘要

PROJECT SUMMARY/ABSTRACT The metabolite and neurotransmitter profiles of neural tissues provide a unique window into brain’s physiological state and can be used to extract potential biomarkers for detecting and characterizing neurodegenerative diseases. Magnetic resonance spectroscopic imaging (MRSI) allows simultaneous mapping and quantification of a number of metabolites and neurotransmitters without exogenous contrast agents thus promised tremendous opportunities for molecular imaging of the brain. However, due to several fundamental technical challenges, including low SNR, poor spatial resolution, long imaging time and inaccurate separation of spectrally overlapping molecular signals, most in vivo MRSI studies to date are still limited to very low-resolution experiments (~1cm3 voxel size) with small brain coverages. The primary goal of this proposed research is to develop, optimize and evaluate a new framework to model, acquire and process MRSI data to enable simultaneous, high-resolution, whole- brain mapping of metabolites and neurotransmitters in clinically feasible time. To achieve this goal, in Aim 1, we will design and implement a novel acquisition strategy that synergistically combines SNR- efficient, multi-slab and multi-TE excitation, sparse sampling in a (k,t,TE)-space and optimized TE selection with maximum echo sampling to generate J-resolved (multi-TE) MRSI data with an unprecedented combination of speed, resolution and organ coverage. In Aim 2, we will develop novel nonlinear low-dimensional models of general MR spectra using a learning-based strategy that integrates the biochemical priors of neural tissues, known physics-based MRSI signal modeling and deep neural networks. These learned models will effectively reduce the dimensionality of the imaging problem and allow for significantly improved speed, resolution and SNR tradeoffs as well as signal separation. Novel computational solutions that effectively exploit the learned models and other spatial-spectral-TE constraints will be developed for spatiospectral reconstruction of metabolites and neurotransmitters from the noisy, high-resolution J-resolved MRSI data. Finally, in Aim 3, we will systematically evaluate the proposed technology in terms of speed, resolution, SNR, and quantitative accuracy using computer simulations, phantom and in vivo experiments. The feasibility and robustness of the proposed technology for mapping metabolites and neurotransmitters in both healthy volunteers and temporal lobe epilepsy patients with mesial temporal sclerosis will be demonstrated. The success of the proposed research will lead to significant progress for in vivo MRSI and represent an important step towards the creation of a powerful tool for studying the molecular basis of brain functions and diseases. This tool, when fully developed, will add a transformative dimension to the existing neuroimaging technology profiles, with the potential to impact the diagnosis and management of neurological and neurodegenerative diseases.