Hi,
I have recently come across an issue with running fmriprep on a large dataset, roughly half of the files have pre-processed functional scans that have part of the brain cropped out (specifically the posterior aspect of the cerebellum).
Looking back through the .html file, I found that during the alignment of functional and structural MRI data, the transformation of the EPI-space to T1w, appears to be outlining regions of the brain that are then cut out from the functional data. Even looking the generated brain mask, outlines the same cropped brain, it appears as if the pipeline in fmriprep is misidentifying tissue, and removing it.
This has resulted in pre-processed functional data that has part of the cerebellum missing. Is there a way to prevent this from occurring in fmriprep?
Only possible explanation that I could consider causing this issue is because in some of these MRI scans, patient’s head was relatively titled and the entire brain for fmri sequence just fitted into the frame, not sure if this is why fmriprep is cropping out the lower aspect of the brain.
Appreciate anyone’s feedback or advice.
Natasha