A whole brain one-way ANOVA was used to examine the between-group differences in both FA and MD on a voxel-by-voxel basis. Type I error was controlled using intensity and cluster based thresholding (family-wise p<.05). A Monte Carlo simulation with a alpha =.05 and connectivity of 1 mm [85] determined that clusters comprised of at least 27 contiguous voxels would have a 5% chance of occurring under the null hypotheses. Thus, only clusters consisting of ≥27 microliters with each voxel differing at p<.05 were interpreted. Follow-up analyses determined which groups were responsible for significant ANOVAs.
