Do macromolecular and spline baselines affect the metabolite quantification at 9.4T?

Purpose/Introduction: One of the major goals in order to make MRSpectroscopy relevant for clinical practice is to ensure a robust quantitation of the spectra. Several previous studies investigated the influence of the macromolecular baseline(MMB) on the spectral quantification accuracy and precision1–12. This work extends previous work by investigating the influence of (1) the brain region specificity of the macromolecular baseline model and (2) the parameters of the additional spline-baseline on the quantification accuracy of spectra measured in the human brain at 9.4T. Subjects and Methods: Metabolite-cycled semiLASER (MCsLASER) 13 spectra (TR = 6000 ms/TE = 24 ms) were acquired from 8 volunteers(aged 28 ± 3) at a 9.4T Siemens Magnetom scanner from the occipital lobe(Occ.) with a mixed grey(40) and white matter(56) content, and from the left parietal lobe (lPar.) with a high white matter content(80). MMB models were created from data acquired using a double inversion MCsLASER sequence(-TI1 = 2360 ms/TI2 = 625 ms)14 from both regions (Fig. 1). using either of the two different MMB models to fit spectra from both brain regions; (2) varying parameters of the spline-baseline stiffness. The paired Wilcoxon test was used to find statistically significant differences in metabolite concentrations. Results: Up to 16 metabolites were readily quantified for both regions with Cramer Rao Lower Bounds of less than 20. The changes in concentrations for quantifying the spectra using the MMB from the corresponding region, versus using the MMB from the other region showed no significant difference (Fig. 2). The parameter settings for the spline-baseline did not influence this result. A-B show the fitted metabolite concentrations and their standard deviations for the different subjects, comparing the use of the different MMB for the fitting of the left parietal and the occipital lobe respectively (dkntmn = 1). The lowest measured p-value for a metabolite concentration change upon changing the MMB was 0.093 (not corrected for multiple comparisons). C-D show sample fit results for using the different MMBs for a left parietal spectrum. However, changes in the stiffness of the fitted spline-baseline lead to statistically significant concentration differences for multiple metabolites between stiff and more flexible spline-baselines (Fig. 3). Already moderately flexible spline-baselines (dkntmn\0.5) influence the accuracy and precision of concentration estimates for several metabolites. A–B show the fitted metabolite concentrations and their standard deviations for the different subjects, comparing the use of the different stiffness parameters of the spline baseline for the fitting (the region specific MMB was used). The red stars indicate metabolite concentration pairs for which a significant change in concentrations was observed (p-value\0.05, not corrected for multiple comparisons). C-D show sample fit results for using the different spline baseline stiffness values for an occipital spectrum. Discussion/Conclusion: This work shows for the first time, that using MMBs from different locations (lPar. and Occ.), with different underlying tissue composition does not lead to significant changes in the fitting results of the metabolites at 9.4T human studies. However, it was found, that the stiffness of the additional spline-baseline fit can significantly influence the metabolite concentrations, hence the flexibility of the spline-baseline should be restricted when quantifying spectra