Comparison of two approaches to model the macromolecule spectrum for the quantification of short TE H-1 MRS spectra

Accurate assessment of the macromolecule contributions in short echo-time proton MRS spectra is important for a reliable quantification of the neurochemical profile. The aim of the present study was to assess two approaches based on using LCModel, which take the macromolecule contributions into account in the quantification step at 14.1 T. H-1 spectra were acquired on a 14.1T/26cm system on five rats using the ultra-short-echo time SPECIAL spectroscopy sequence. Metabolite concentrations were estimated using LCModel, combined with a simulated basis-set of metabolites using published spectral parameters and: 1) the spectrum of macromolecules measured in vivo using an inversion recovery technique; and 2) the built-in LCModel spline baseline. The fitted splines resulted in a smooth approximation of the in vivo macromolecules, which could not reproduce completely all features of the in vivo spectrum of macromolecules at 14.1T. We can conclude that, as in previous studies using Subtract-QUEST, even if the fitted splines reproduce the in vivo macromolecules very well, the measured macromolecular "baseline" represents an additional prior knowledge and lead to a more accurate and reliable quantification at higher field strengths, which was attributed to an effectively increased spectral resolution of the macromolecule components.