Longitudinal absolute metabolite quantification of white and gray matter regions in healthy controls using proton MR spectroscopic imaging

Abstract

The purpose of this study was to evaluate quality parameters, metabolite concentrations and concentration ratios, and to investigate the reproducibility of quantitative proton magnetic resonance spectroscopic imaging ((1)H-MRSI) of selected white and gray matter regions of healthy adults. 2D-quantitative short-TE (1)H-MRSI spectra were obtained at 1.5T from the healthy human brain. Subjects (n = 12) were scanned twice with an interval of six months. Absolute metabolite concentrations were obtained based on coil loading, taking into account differences in sensitivity of the phased-array head coil. Spectral quality parameters, absolute metabolite concentrations, concentration ratios, and their reproducibility were determined and compared between time-points using a repeated measures general linear model. The quality of the spectra of selected brain areas was good, as determined by a mean spectral linewidth between 4.8 and 7.3 Hz (depending on the region). No significant differences between the two time-points were observed for spectral quality, concentrations, or concentration ratios. The mean intrasubject coefficient of variation (CoV) varied between 4.0 and 8.5% for total N-acetylaspartate, 7.2 and 10.8% for total creatine, 5.9 and 9.8% for myo-inositol, and 8.0 and 13.3% for choline, and remained below 20% for glutamate. CoV was generally lower when concentration ratios were considered. The study shows that longitudinal quantitative short-TE (1)H-MRSI generates reproducible absolute metabolite concentrations in healthy human white and gray matter. This may serve as a background for longitudinal clinical studies in adult patients.

Influence of macromolecule baseline on 1H MR spectroscopic imaging reproducibility

“…Table 6 shows the absolute concentrations for the metabolites taken from white matter for each TE and a comparison between Si.BL and Ex.BL analysis schemes. For TEs = 35 and 80 ms, absolute concentrations were found to be consistent with those found in literature for healthy parietal white matter 19, 29. However, a slight underestimation was observed for TE = 144 ms suggesting accuracy may be reduced at long TEs due to increased T2 relaxation.…”

“…This study mainly focuses on the intrasubject reproducibility within the same session rather than accuracy. Metabolite concentrations were calculated for each TE and MM analysis scheme and were found to be consistent with those previously published in literature for parietal white matter 19, 29. A slight underestimation was found for a TE = 144 ms, suggesting accuracy is decreased at longer TE, which could be due to T2 bias on the metabolite estimates.…”

Reproducibility of quantitative structural and physiological MRI measurements

“…In 10 healthy subjects scanned twice, Jansen and colleagues noted metabolic MCVs of 7.0%–20.4% and ICCs of 0.00–0.55 in the frontal and temporal lobes (Jansen et al., 2007). Wiebenga and colleagues noted a slightly higher MCV in 12 subjects with 6 months between TE30 scans (Wiebenga et al., 2014). The intrasubject higher MCVs reported by Ding and colleagues reflect his whole‐brain reproducibility metric rather than the small region of interest used in our study (Ding et al., 2015).…”

Enhanced Axonal Metabolism during Early Natalizumab Treatment in Relapsing-Remitting Multiple Sclerosis

“…17 Although the chemical shift artifacts are not very large at 1.5T, there are distinct differences between the first and last rows. They were therefore discarded in the analysis to avoid additional variation due to this effect ( Fig 1A).…”

“…2D-MRSI included a point-resolved spectroscopy sequence (TR/TE, 3000/30 ms) on a single 15-mm slab aligned to the sections of the proton-attenuation/T2 sequence, with the center touching the top of the corpus callosum. 17 Depending on head size, the FOV was 160 ϫ 160 or 140 ϫ 160 mm and the corresponding volume of interest was 70 ϫ 100 or 80 ϫ 100 mm. The use of 16 ϫ 16 phase-encodings resulted in a voxel size of 1.3 or 1.5 mL.…”