Hi,
 
I'm sitting on a bunch of data that requires quality control (QC). While there are several guidelines for skullstrip or gray/white matter segmentation QC, I couldn't find anything on subcortical QC. As far as I know freesurfer classifies the different structures probabilistically using a preexisting atlas. This makes it rather impossible to QC the segmentation purely based on intensity differences. 
So I was asking myself if there might be any additional possibility than to merely rate the quality of the subcortical segmentation by more or less subjectively comparing their forms with a hypothetical ideal form?
 
Thanks,
 
Jurij