Hi all,
I'm doing a tbss analysis and looking for a correlation between white-matter FA and a physiological variable. It's very interesting/suspicious that the statistic is significant for about half of the tested skeleton-voxels using TFCE, while there is only 1 significant voxel using voxel-wise statistic (10000 permutations and corrected p-values in each case). I know how TFCE works, but it is very unlikely that the found results are so different using voxel-wise and TFCE tests and that our physiological variable affects the FA in so many skeleton voxels as found by TFCE. How could I clarify which test should I trust? The registration and the quality of DTI images seem to be good enough.
Thanks,
Gábor
|