MRS fingerprinting combined with multi-dimensional linear-combination modeling for fast and accurate multi-metabolite relaxometry.

Motivation: Metabolite relaxation times are required for quantification, but rarely estimated for every subject due to time constraints. Therefore, time-efficient MRS methods for simultaneous metabolite concentration and relaxation time estimation are highly desirable. Goal(s): Improve time-efficient estimation of subject-specific metabolite relaxation times. Approach: Magnetic resonance spectroscopy fingerprinting (MRSF) is combined with multi-dimensional linear-combination modeling (LCM) for metabolite relaxation time estimation. The performance of conventional MRSF and MRSF combine with LCM is evaluated with Monte-Carlo simulations. Results: The combined method estimates relaxation times for five major metabolites (NAA2.0, Cr3.0, Cho, mI, Glx), achieving higher accuracy than conventional MRSF, but lower precision. Impact: MRSF combined with linear-combinaiton modeling enables accurate cohort-mean relaxation time estimation for five metabolites suitable for quantification in large studies, while conventional MRSF (using simple peak integration) may better detect small subject-specific changes in singlet relaxation times.