3dSkullStrip over-stripping top of brain

Hello AFNI community,

I am having trouble applying AFNI’s 3dSkullStrip to the anatomical data from the Consortium for Neuropsychiatric Phenomics (CNP) dataset (https://openfmri.org/dataset/ds000030/). Specifically, 3dSkullStrip is consistently overstripping the top of the brain (e.g. regions near the left postcentral gyrus) but otherwise performing well on the rest of the brain.

I have tried many of the recommended approaches in the documentation for 3dSkullStrip, including first running 3dUnifize then trying the following options:

In general, the above options result in too liberal an expansion of the brain region, capturing too much skull in addition to the problem regions at the top of the brain. Ideally, I would want to only change the brain expansion near the top of the brain only and not change it anywhere else.

Any help would be greatly appreciated.

  • Levon Demirdjian