I realized few ROIs defined in TT_desai_dd_mpm+tlrc and TT_desai_dk_mpm+tlrc did not include any voxel!
(I got a warning message, *+ WARNING: output sub-brick 0 is all zeros!)
For probablistic versions of these atlases, voxels with non-zero probability seems to overlap among the regions.
I expected all the ROIs defined in the atlas should have at lest one non-zero (1) voxel.
In the list (http://www.neuro.mcw.edu/~rhdesai/mcw-www/Probabilistic_Atlas.html), few regions correspond to ventricles…
which makes sense. However, in some “normal” regions, there were no voxels for the ROIs.
Do you have any idea on this problem?