Fast High-Resolution Deuterium MR Spectroscopic Imaging Integrating Low-Rank and Generalized Series Modelling

Motivation: Deuterium MRS imaging (DMRSI) enables noninvasive metabolic imaging but suffers from low sensitivity, thus, poor resolution due to the low gyromagnetic ratio of deuterium and low deuterated metabolite concentrations. Goal(s): To develop a method for fast, high-resolution DMRSI that improves spatial resolution and SNR. Approach: We proposed a novel data acquisition scheme with variable-density k-space sampling and an image reconstruction scheme that integrates both spectral and spatial priors using a generalized series (GS)-constrained subspace model. Results: Simulations, phantom studies, and in vivo experiments showed that the method achieved higher resolution, SNR and reproducibility compared to traditional approaches. Impact: The proposed method improves reliability and resolution for imaging human brain glucose metabolism. The method would prove useful for study of metabolic reprogramming in both healthy and diseased states, particularly, for mapping brain tumor Warburg effects and assessing intra-tumor heterogeneity.