Test-retest reliability of diffusion measures in cerebral white matter: A multiband diffusion MRI study.

Abstract

To investigate the test-retest (TRT) reliability of the diffusion measures in cerebral white matter obtained from the diffusion MRI dataset acquired with multiband acceleration. With the multiband diffusion MRI dataset with two repeated scanning sessions, the TRT reliability of diffusion measures (fractional anisotropy [FA], mean diffusivity [MD], primary diffusivity [PD] and transverse diffusivity [TD]) was investigated through several fully automated analysis methods, including two voxel-level analyses (voxel-based analysis [VBA] and tract-based spatial statistics [TBSS]) and an atlas ROI-based analysis. The reproducibility was assessed by the intra-class correlation coefficient (ICC). Our results demonstrated moderate to high reproducibility (ICC > 0.4) of diffusion measures from the multiband EPI sequence with different analysis approaches. Across different measures, FA exhibited the highest reproducibility (mean ICC = 0.70), while MD showed the lowest reliability (mean ICC = 0.55) (P = 0.006). Additionally, ICCs varied across different tract ROIs: Commissural tracts showed higher reproducibility than other categories of tracts (projection, association and brainstem), while the brainstem tracts exhibited the poorest reliability (P = 0.004). Our findings suggest a potential utility of the multiband EPI sequence for exploring individual differences of cerebral white matter and provide reference for future white matter studies.