I'm currently trying to take manually made ROIs (label files) that have been incorporated into new annotation files (one for each hemisphere - rh.aug_aparc.annot & lh.aug_aparc.annot) and turn them into a new (augmented) annotation into a aseg file (an aug_aparc+aseg.mgz file).

To do so I've been using the mri_aparc2aseg command and, at face value, it works well. However, some of my ROIs (specifically my insular ROIs) are being either shifted or lost in the transition from annotation to segmentation. Note that each hemisphere's new annotation files look perfect on tksurfer (which is where they were originally drawn). It's only when I try and make them fill the full cortical ribbon with the mri_aparc2aseg command that I get this issue.

I piloted everything out on averaged subject and it all worked perfectly. However, when I attempted things on individual subject data I received the problem described above. I've attached a picture illustrating the discrepancy between the an averaged subject and a single subject. I have also attached my script code.

The overall goal is to create a gmroi_volume which will allow me to determine an aux number of each manual ROI so as to export each one to AFNI space.

I'd really appreciate help on how to fix this.

Thanks,

Paul