We are having trouble with getting the eye area removed. We have tried playing with shrink factor up to .9. While that helps clean up things near brain stem, we still have eyes present on many scans.
Do you have a some 3dSkullstrip parameters that might work for this sort of problem?
If you are also transforming to MNI space, you could try the script @SSwarper. It is slow, though.
Paul points out I was somewhat terse in my reply last night. The outputs of this script are described when you type
or when you follow this link: https://afni.nimh.nih.gov/pub/dist/doc/program_help/@SSwarper.html
The output files include the skull-stripped dataset in original coordinates, and as warped to the MNI 2009 template dataset. The transformation files needed for afni_proc.py to avoid re-doing the nonlinear warping are also save. The “-help” output also describes how these files are named and how to input those files to afni_proc.py.
Thanks for the suggestion. I tried @SSwarper and the processing did not fully finish. Seemed to get stuck on my iMac and got a broken pipe error on a MacPro. However, I think the final skullstrip image was created and that does indeed look much better (though some cleanup near midbrain could be added). How much memory and processor speed is necessary for @SSwarper? If this works well, I’d like to use it as my default if it actually completes running.
Do you have a newer iMac? That should be good, if you have enough memory. I’d say 16 GB, considering the OS (etc.) takes about 6-8 GB. If you run on a laptop, anything using 3dQwarp (such as @SSwarper) will take a long time.
I run this on my MacPro all the time, and have never seen a pipe error. However, I have 32 GB of RAM.
I’ll have to do some checking. The MacPro I ran on also has 32 GB of RAM. How long should one job take?
30-60 minutes, as I recall (it’s been a few weeks since I did this).