Hello,

I am preparing a dataset of 42 segmented vervet brains  to make an aseg atlas using mri_ca_train.  A couple of quick questions:

1. Does every voxel need to be labeled for mri_ca_train to create a robust atlas? E.g., we have not manually segmented out white matter or cerebral cortex in our brains.  Is this necessary?

2. Can I create a .gca file with mri_ca_train that can then be used by mri_ca_label to label a novel dataset without having to use rebuild_gca_atlas.csh?

Thanks!

Kevin Scheibel

k.scheibel@ucla.edu