Abstract
We investigated test-retest reliability of two MRI-derived indices of white-matter microstructural properties in the human corpus callosum (CC): myelin water fraction (MWF) and geometric mean T(2) relaxation time of intra/extracellular water (geomT(2IEW)), using a 3D gradient and multi spin-echo sequence in 20 healthy adults (aged 24-69 years, 10 men). For each person, we acquired two back-to-back acquisitions in a single session, and the third after a break and repositioning the participant in the scanner. We assessed the contribution of session-related variance to reliability, using intra-class effect decomposition (ICED) while comparing two CC parcellation schemes that divided the CC into five and ten regions. We found high construct-level reliability of MWF and geomT(2IEW) in all regions of both schemes, except the posterior body-a slender region with a smaller number of large myelinated fibers. Only in that region, we observed significant session-specific variance in the MWF, interpreted as an effect of repositioning in the scanner. The geomT(2IEW) demonstrated higher reliability than MWF across both parcellation schemes and all CC regions. Thus, in both CC parcellation approaches, MWF and geomT(2IEW) have good test-retest reliability and are, therefore, suitable for longitudinal investigations in healthy adults. However, the five-region scheme appears more appropriate for MWF, whereas both schemes are suitable for geomT(2IEW) studies. Given the lower reliability in the posterior body, which may reflect sensitivity to the repositioning of the participant in the scanner, caution should be exercised in interpreting differential findings in that region.