I am currently attempting to run recon-all on infant subjects with a newly created 'gca' atlas.  I e-mailed the listserv a while back about a strange memory allocation error that occurred right at the beginning of the "subcortical segmentation" pipeline. For a while, we thought that it was an issue with the actual atlas. However, after talking with someone who created canine atlases, it seems like it is a particular bug in the recon-all scripts themselves.

       It appears as though the default recon-all script needs to be altered to accept the new gca atlas. I was wondering if anyone has had success getting Freesurfer to accept their gca atlas? Thank you!

Best,