Hello Jim Bjork:
I have an approach to do what you want that utilizes the program 3dROIstats and its option -voxcount.
1) Use 3dmerge to threshold each individual subject's data. I presume you have individual subject contrasts for each of the contrasts you listed for your group map, so you would want to threshold these with the same threshold (t>3.3).
2) Use the 3dmerge options -1clust_order 1.09 6 and -1thresh 3.3 on your group map so that each cluster in the brain is assigned a unique number.
3) Use 3dcalc to create a mask out of only those clusters present in the OFC. So assuming your OFC clusters are #3 and #5:
3dcalc -expr '(1-bool(a-3))*((1-bool(a-5))*2)' -a groupmap_clustorder_thresholded+tlrc -prefix OFC_clusters_mask
Bool turns TRUE values to 1, so cluster #3 will now have a value of 1 and cluster #5 will have a value of 2.
4) For each contrast, use 3dROIStats to sample each subject's thresholded data with your mask.
foreach subject (subj1 subj2 subj3)
3dROIstats -voxcount -mask OFC_clusters_mask+tlrc ${subject}.thresholded_contrast1+tlrc >>OFC_voxcount
end
this option counts the number of active voxels within each OFC cluster for each subject. You can add them together to get an overall OFC voxelcount. Additionally, if you get the OFC cluster numbers from your clust_order file you can also do 3dROIstats on the clust_order file and just use the voxcounts from the columns for clusters #3 and #5. There really isnt a need to extract only the OFC clusters first other than it saving some processing time.
Hope this helps,
Christine Smith
