Voxelwise Analysis of FA Data: Session and Subject Variability, Gaussianity and Dependence on Acquisition
Smith S, Johansen-Berg H, Mackay C, Bartsch A, Behrens T
We have recently proposed a new approach to voxelwise analysis of FA data: Tract-Based Spatial Statistics (TBSS). TBSS pre-aligns all subject? FA data, creates a mean-FA-based white-matter skeleton image, and then projects each subjec?s FA image onto this skeleton, before carrying out cross-subject voxelwise analysis. Here we investigate: a) the cross-subject Gaussianity of FA data; b) the cross-session and cross-subject variability in FA when pre-processed by pure nonlinear registration, TBSS and by hand; and c) the dependence of cross-subject FA variability on different DTI acquisitions (6 vs. 60 directions and single-channel vs. parallel imaging).