Comparison of two strategies of background-accommodation: Influence on the metabolite concentration estimation from in vivo Magnetic Resonance Spectroscopy data

Abstract

Localized proton Magnetic Resonance Spectroscopy brain signals acquired at short echo-time contain contributions from metabolites, water and a ;background' which mainly originates from macromolecules and lipids. The purpose of the present study was to compare the influence of the background-accommodation strategy on the metabolite concentration estimates. Two strategies were investigated to accommodate the background, 1) the measured background signal was incorporated in the metabolite basis-set; and 2) the background signal was estimated and subtracted from the in vivo signal using Subtract-QUEST. The influence of the background-accommodation strategy was addressed with the aid of Monte Carlo and in vivo studies. For the considered signals of this study, the concentration estimates obtained using the first approach were below those obtained with Subtract-QUEST. Indeed, the presence of residual contribution of metabolite signals with short longitudinal relaxation times (T1) in the measured background led to an underestimation of metabolite concentration estimates. Conversely, the observed underestimation of the background contribution using Subtract-QUEST led to an overestimation of the metabolite estimates.