I was doing QA on the preprocessed data, and noticed that some fMRI data is outside of the anatomically defined outlines (shown in the screenshot below). It is also part of the brain mask generated by fmriprep.
I was wondering if the outside part is indeed non-brain artifacts or is due to some registration error? And how should I treat it in the statistical analysis? Thanks!
The coregistration looks good - what you are seeing is a susceptibility distortion artefact which is very common. This can be corrected with additional set of scans (called field maps) or if you don’t have those you can try the
--use-syn-sdc FMRIPREP flag.