Dear experts,
a question was raised whether it is possible to automatically segment anterior and posterior commissures. Do you think that it is achievable with reasonable accuracy, using in-vivo data of 0.7 mm3 resolution?
I was thinking what would be the best way: a) Is it possible to customize Hippocampal subfields segmentation framework to achieve this goal?
b) I considered the most straightforward option to extend current RB_all_2016-05-10.vc700.gca atlas to include also labels of anterior and posterior commisure. However, this would need to have labeled source subjects for this atlas. Is it publicly available? If yes, do you think that it makes sense to do it this way?
c) In case of source subjects for RB_all_2016-05-10.vc700.gca are not available, then there would be option to create the atlas from scratch. Does it make sense to do it like following: Take let's say 10 representative subjects, run them through recon-all, take their aseg.mgz, manually label their commissures (only, other stuff in aseg.mgz leave as is) and run rebuild_gca_atlas.sh on relabeled aseg.mgz? This (in contrast to do the manual labeling of whole brain) seems doable to me. I do not know how the resolution of subjects constituting RB_all_2016-05-10.vc700.gca was, so maybe doing labeling on our 0.7mm3 subjects could lead to even more accurate results, than option b)? Could you please comment on?
Regards,
Antonin Skoch