Two pre-alignment processing steps were conducted: a six-degree of freedom affine motion correction for head motion; and a 2D, six degree of freedom alignment to reduce the effects of gradient coil eddy currents (FLIRT) [44]. Each image was inspected for quality, and non-brain voxels were removed from analysis by AFNI 3dAutomask, then manually refined as needed. A representative diffusion image from the control group was selected as the target image for registration. All subjects were aligned by nonlinear transformation to the selected target image and then the whole dataset was affine-aligned to MNI-152 space [30]. These transformed data images (FA and MD) were averaged to create a mean FA or MD image. Next, a white matter tract skeleton was created from the mean FA or MD image to represent the centers of large white matter tracts in the dataset. This skeleton was generated by identifying lines (within white matter tracts) in the mean image with the largest diffusion values. Finally, each participant's transformed FA and MD image was projected onto the diffusion skeleton. This procedure helped achieve alignment between the
