Magnetic Resonance Spectroscopy Signal Analysis Based on Fingerprinting Dictionary Approaches

Abstract

Magnetic resonance spectroscopy (MRS) is a technique applicable in medical diagnosis or research, which has the unique capability to give non-invasive access to the biochemical content (metabolites) of scanned organs. Up to recent times, all the proposed methods solved metabolite quantification as an optimization problem attempting to minimize the difference between the data and a given parameterized model function. This paper proposes quantification of metabolites in MR spectroscopic imaging using a fingerprinting method, whose function is based on the creation of a dictionary of linear combinations of metabolite signals. Experimental results demonstrate the accuracy of the proposed method, compared to data obtained by a standard quantification method (QUEST), on concentration estimates of 8 metabolites from signals with macromolecule background and noise. The prototype results indicate that the concept of MR fingerprinting dictionary, useful also for preparing data for machine learning, can serve as an alternative method for metabolite quantification by NMR signal analysis.