Thanks for a prompt reply. OASIS tempalte was used. If we padded the image with a couple of layers of zeros prior to running brain extraction that could help?
More context and raw data here: https://neurostars.org/t/what-to-do-when-fmriprep-goes-wrong/508
I know input data is bad, but I was wondering if anyone would have any suggestions about options or preprocessing steps that would make ANTs brain extraction more robust to such cases. Thanks in advance!
It's been over 2 years since version 2.1.0 was released. Myriads of fixes and new features were introduced to ANTs since then, Maybe it's time for version 2.2.0? Having numbered releases helps to track what exact code was used in a particular data processing and thus improves transparency and reproducibility.
We have a workflow for this in Nipype. Works on SGE, Condor and others: http://nipy.org/nipype/users/examples/smri_antsregistration_build_template.html...
I saw it in MIPAV as well.
The delineation I did was quite accurate, but I blame this on my OCD ;) I did not...