Dear Mike,
The easiest approach is to go with an absolute threshold for masking during model specification, something like e.g. 0.1 or 0.2. This way only those voxels will become part of the analysis whose intensity in the specified files (usually some smoothed version) is larger than the specified value for every single subject.
If you want to generate an average GM image for display purpose you could average the different images (e.g. the swm* ones) and mask them with the mask image obtained during model estimation.
This paper http://www.ncbi.nlm.nih.gov/pubmed/18848632 might also be interesting when it comes to intensity thresholds for VBM analyses.
Best
Helmut
|