I've added a new feature to the FreeSurfer Group Descriptor (FSGD) file which allows you to easily use the volume subcortical structures as covariates in a mri_glmfit analysis. An example is:
GroupDescriptorFile 1 Title test
Class main
VariableFromASeg Left-Hippocampus Variables Age
Input subject1 main 20 Input subject2 main 30 Input subject3 main 40 ...
The above fsgdf will have two variables: Age and the volume of the left hippocampus. The volume is obtained from stats/aseg.stats. Within the design matrix, the left hippocampal regressor will appear after that of Age (you'll need to know this to set up your contrast matrices). mri_glmfit will fill in the values. mri_glmfit will write an fsgd as part of it's output, and this will include the volumes.
This is formally equivalent to something like:
GroupDescriptorFile 1 Title test
Class main
Variables Age Left-Hippocampus
Input subject1 main 20 2000 Input subject2 main 30 3000 Input subject3 main 40 4000 ...
You can have more than one VariableFromASeg statement.
There are mechanisms to use something other than aseg.stats (eg, if you have your own continuous variables such as age, IQ, etc).
I wrote and tested this on a flight from new mexico to boston, so I have not tested it out extensively. Let me know if you experience any turbulence.
I'll add some docs to the wiki when I have time.
doug
freesurfer@nmr.mgh.harvard.edu