When I run 3dAutomask on about half of my subjects, I clip off a lot of low intensity voxels in the temporal and orbitofrontal lobes, including part of the amygdala - where I really want to see activation.

When I run 3dAutomask with -dilate 3 as a option - the resulting dataset includes those missing voxels - but it also ends up adding a lot of non-brain voxels particularly around the top of my head. I am guessing having those non-brain voxels will interfere with scaling the dataset.

So my question is, can I save those low intensity voxels but not include too much non-brain in my masks?

During Bootcamp Ziad had mentioned adding in a d term to 3dcalc that would discard voxels under a certain threshold - but I can't figure out how to set that up properely. I am guessing the threshold value would change from subject to subject - maybe even run to run.

Just wanted to see what options are out there.

Thanks so much!

Tracy Jill