Hi Liana,
The "number of voxels" problem is surely because of resolution
differences between the datasets. Consider "3dresample -master ...".
As Isaac noted, those voxels are outside the mask of all subjects.
Actually, 94377 is not a big number, for voxels that are all zero across
your subject masks. The 3 mm^3 EPI mask that I am working with has about
270000 voxels, about 200000 voxels of which are all zero. I would expect
similar numbers per subject, and probably 180000 or so that would be zero
across all subjects (had I masked at the individual level). For a 3 mm
case your 94000 seems small.
But the voxel count depends on the voxel size.
- rick