Hi,
Sorry, this may not make much sense.
I am doing some standard resting state functional connectivity processing.
I was using 3dAutomask on the epi to remove the non-brain areas.
For some subjects, there is some signal loss at the bottom of the brain (as expected), and the coverage mask looked fine.
Upon closer inspection, there does seem to be some "real" data just one voxel out from the edge of the mask.
So I dilated the mask by one voxel (epi space) and reprocessed.
Issue 1: Once everything is in MNI space, the edges of the dropout regions seem to be expanded more than edges elsewhere, like the top of the head.
Issue 2: when I look closely at the data at the edges, it seems like the processed dilated image has higher intensity at the edges than the non-dilated images.
For example, I put the dilated image as the underlay and the non-dilated as the overlay.
At the edge where they do overlap, the dilate image has a higher intensity than the non dilated.
If I go further in away from the edges, the intensities are similar.
Long story short...
How do I know if there is "real" data at the edges of the drop out?
and is it possible to quantify it?
Thanks, Justin