Hi Daniel, I'm reposting this to the FS list because I don't really use mri_ca_train. doug

On 02/26/2013 02:45 PM, Barron, Daniel S wrote:

Hello Dr. Greve,

I am trying to create a .gca for thalamic nuclei. Based on the Morel 2007 atlas, I have created 6 nuclear grouping masks per hemisphere. I created these masks in MNI space with fslmaths. Each mask is a separate, binary .nii.gz.

In preparation for mri_ca_train, I have converted the .nii.gz files to .label files with mri_cor2label. However, I am unable to assign vertex numbers for each voxel (because the thalamic volumes do not correspond to a surface?). When I transform the binary .nii.gz-->label files, in each .label the entire vertex number column reads "-1". I have attached one such label as .label.txt. This carries over even after I register each to FreeSurfer space (mri_label2label). The problem is when I go to create an annotation file, I get an error saying the vertex number is -1. Can you recommend a way for me to navigate this error?

Once I have created the annotation file, can I use mri_aparc2aseg to create the segmentation files for mri_ca_train? mri_aparc2aseg --help says this is intended for surface annotations; I want to use it for volumetric annotation.

Thank you very much for your help~

Cheers, Daniel