Hi Freesurfers,
I am interested in creating new atlas for Freesurfer reconstruction, and I am having some trouble figuring out where to start and where to edit my labels though I have been researching it for some time. I have emailed Mark Plantz, who has been asking you similar questions with UNC's 1yr AAL atlas, but I am not sure if he has gotten it to work yet. I have looked at the documentation, but it is unclear to me.
I have the following materials: 100 or so age-specific infant brains that do not segment, reconstruct, nor brain extract very well in Freesurfer due to the immature nature of the brains. I have a nice average diffeomorphically warped template, 116-region version of the AAL atlas which applies to this template we made, and segmentation information, all in NIFTI format and RAS coordinates.
I want to do the following two things: 1. Make preliminary GCA non-probabilistic atlas using 116-region AAL atlas. 2. Make GCA probablistic edits using manual edits of the individual brains. 3. Set freesurfer to use these atlases for recon, infant brain size information for future brain extraction, infant segmentation information for future. Have freesurfer perform the most accurate possible WM segs on all our infants in the future for purposes of MEG localization.
My questions are: 1. What exact steps do I need to do to my original AAL.mgz volumteric atlas to get it ready to be converted to surface format and gca and mr_ca_train? combined with my new Lookuptable I created? I have looked at all documentation and this remains elusive to me. 2. Do I need to Freesurfer all brains I want probability information from, or can I hand segment in another program, and then use these segmentations as a training set? 3. What are all these atlases, such as RB_all_2008-03-26.gca RB_all_withskull_2008-03-26.gca etc in the pipeline and can I change them to be my own infant information to improve skull stripping with the --atlas, etc?
Thanks in advance, Liv Wroblewski