Multicenter Measurements of Myelin Water Fraction and Geometric Mean T2: Infra- and Intersite Reproducibility

PurposeTo assess the reproducibility of myelin water fraction (MWF) and geometric mean T-2 (GMT(2)), which are in vivo markers of pathological changes underlying disability and progression in diseases such as multiple sclerosis. Materials and MethodsFive healthy volunteers were scanned twice within 24 hours at six different sites using the same manufacturer's 3T magnetic resonance (MR) system. T-2 distributions were produced by fitting multiecho 3D T-2 data using non-negative least squares, with stimulated echo correction. MWF, the fraction of signal with T-2 between 15 and 40 msec to the entire signal, and GMT(2), the mean T-2 on a logarithmic scale from T-2 between 40 and 200 msec, were examined in white matter. ResultsIntrasite coefficients of variation (COVs) were low (mean 3.99% for MWF and 0.51% for GMT(2)), as were intersite COVs (mean 4.68% for MWF, 0.31% for GMT(2)). Scan-rescan intraclass correlation coefficients (ICCs) (0.76 for MWF and 0.93 for GMT(2)) and Bland-Altman plots indicated good agreement between single site scans. Intersite ICCs were relatively high (0.69 for MWF and 0.92 for GMT(2)), revealing good intersite reliability. ConclusionMWF and GMT(2) measures are reproducible between scans and across sites with an equivalent MR scanner and sequence protocol. Multicenter clinical trials using quantitative T-2 relaxation are feasible.