I finished processing my first longitudinal subjects, but I have some errors. The TP3.long aseg.mgz file is poorly segmented (corpus callosum, see attachment). I am wondering where I should make edits to correct this. The surfaces of all 3 longitudinal timepoints look fine, so according to the wiki, I should only have to edit the TP3.long. Is that true? If not, then I will have to edit TP3.cross. Originally, the watershed for TP3.cross was bad. I edited this and re-ran autorecon2 and autorecon3, and exited with no errors. However, the cross aseg.mgz is still bad. I thought maybe the talairach.m3z file was bad, but the nu_noneck image was transformed fine. I see that the log file says during the mri_ca_label step that the CSF peak is too bright - rejecting (see attachment). After comparing the csf areas of the TP3.cross norm image with the csf areas of the TP1.cross norm image, they have values of 30 compared to values of 10. Previous posts have suggested to copy the orig.mgz image to the nu.mgz image for a similar problem. Can someone please advise me on where it would be best to make the edits so that my longitudinal reconstructions are accurate?