I am still having the same issue as in my previous post. I have successfully run dt_recon and upon looking at the images there seems to be no issue with the registrations of outputs including the FA, lowb and other images. 

I wanted to mask my FA images using the wmparc.mgz image from recon-all using

mri_vol2vol --mov $subj_dir/dtrecon/${subj}.${cond}./lowb.nii.gz --targ $subj_dir/diffusion_recons/${subj}.${cond}/mri/wmparc.mgz --inv --interp nearest --o /Volumes/DANIEL/dti_freesurf/diffusion_recons/${subj}.${cond}/mri/wmparc2diff.mgz --reg $subj_dir/dtrecon/${subj}.${cond}./register.dat --no-save-reg

This works perfectly for all of my subjects except 2 of them. These subjects seem to have normal lowb, wmparc, and FA images compared to the other 58 subjects. However, the wmparc2diff that is created from mri_vol2vol has significantly shifted away from the lowb and fa so that when I try to mask the fa with

mri_mask $subj_dir/dtrecon/${subj}.${cond}./fa.nii.gz /Volumes/DANIEL/dti_freesurf/diffusion_recons/${subj}.${cond}/mri/wmparc2diff.mgz $subj_dir/dtrecon/${subj}.${cond}./fa-masked.nii.gz

It cuts a large part of the brain off and masks incorrectly.

Is there a way to troubleshoot the registration of wmparc to wmparc2diff? I am not sure why the registration is so off? 

Any troubleshooting help for why the registration of the segmentation to diffusion space is so poor (or if the issue is registration from diffusion space to anatomical space) would be greatly appreciated.

Best,