Reproducibility of Quantitative Cerebral T2 Relaxometry, Diffusion Tensor Imaging, and 1H Magnetic Resonance Spectroscopy at 3.0 Tesla

2, slice thickness 20 mm, TR 2.0 s, TE 30 ms, a nominal voxel size of 3.3 ml, echo acquisition half echo. Localization and water suppression was achieved with PRESS and CHESS, respectively. The T2 was calculated (in ms) on a voxel-by-voxel basis using the signal intensities of the images obtained at the two echo times (Matlab). From these values also a percentile volume cerebrospinal fluid (CSF) map was calculated. The ADC (in 10 -6 mm 2 /s) and FA (in %) maps were calculated using the diffusion software available on the MRI scanner. The T2-, CSF-, FA- and ADC-maps were spatially normalized to Talairach space, to facilitated analysis of brain regions with masks. Absolute metabolite quantification was performed using LCModel and the calibration strategy was based on the principle of reciprocity [2]. Concentrations were corrected for cerebrospinal fluid (CSF) contribution, using the CSF-map. The metabolite estimations for choline (Cho), creatine (Cr), myoinositol (mI) and n-acetyl-aspartate (NAA) were analyzed. Metabolite estimates were excluded from reproducibility analysis, if the Cramer-Rao minimum variance exceeded the 20% range. Furthermore, at least 5 volunteers should have reliable spectra in a voxel. Statistical analysis of the T2-, ADC-, and FA-maps was performed in the frontal and temporal lobe, and the entire cerebrum. Descriptive statistics were derived on a voxel-byvoxel basis, and then summarized per region by calculating the median. The coefficient of variation (CV, derived as the mean within-subjects standard deviation (SDws) divided by the mean value for all subjects [3]), the repeatability coefficient (RC = 1.95 * 2 * SDws [3]) and the intraclass correlation coefficient (ICC = SDbs 2 / (SDbs 2 + SDws 2 ), where SDbs