I would just add one more step to Dr. Smith's suggestion. Dr. Nichols suggested the following when I had a similar problem:
Pull up your t-stat map for the results that are significant, and see if they are shifted from 0 (i.e., the peak is not at 0). If this is the case, it suggests that you may have a global RD effect- which you can report. You can then correct for by extracting the global RD value for each subject (using fslmeants -i all_RD_skeletonized -m mean_FA_skeleton_mask -o my_4D_RD_tbss_output).
|