Repeatability and reliability of Diffusion Tensor Imaging in white matter rectified with free-water

Motivation: Recent studies suggest that the free-water (FW) corrected diffusion tensor imaging (DTI) computed at a single-shell not only depends on the intial conditions of the optimisation procedure, but also might be unreliable. Goal(s): To evaluate whether the multiple-shell FW correction enhances the repeatability and reliability of DTI. Approach: The repeatability and reliability were evaluated empirically using the publicly available database MICRA (Neuroimage, 225, 2021). Results: Direct optimization of the so-called bi-tensor representation leads to decreased inter-session repeatability of FW-corrected DTI compared to the standard DTI. Pre-estimating the FW from multiple-shell data and correcting the signal before estimation may improve the repeatability of DTI. Impact: This result for multiple-shell diffusion MRI data opposes previous studies on single-shell data (Hum. Brain Mapp., 38(1), 2017), which suggest that correcting the DTI for FW via direct optimization of bi-tensor representation improves repeatability.