I have one application that needs to run additional freesurfer processing (e.g. brain stem segmentation or hippocampus segmentation) based on the existing fMRIPrep freesurfer results. I am wondering whether I can use the the fMRIPrep singularity container to run these additional processing since I want to keep the preprocessing and processing tools are exactly the same.
Here is the situation:
The additional freesurfer brainstem segmentation can be run based on the existing freesurfer outputs using standalone freesurfer 6.0 with recon-all -s <SUB> -brainstem-structures;
I have finished fMRIPrep 20.2.7 preproc and the freesurfer output folder is ready;
How can I tell fMRIPrep to run only recon-all -s <SUB> -brainstem-structures instead of the standard fMRIPrep and freesurfer anat preproc workflows, do you have any recommendations?
Perhaps you should recon-all outside fMRIPrep, and then rerun fMRIprep; I would not mess with the fMRIPrep source code. If you are worried about version differences, you can do singularity exec -e -B /path/to/important/drives /path/to/fmriprep.img recon-all $arguments.
Great thanks! I was thinking of using freesurfer from the fMRIPrep container directly as well, and I also remembered that fMRIPrep provides some validation functions for these freesurfer outputs, and rerun the specific freesurfer steps if necessary . Do you think running these functions before any further processing will be useful?
I do not know how these functions work, sorry. And now that I think about it, you shouldn’t need to run fmriprep again because it likely doesn’t use brainstem information. So unless the brainstem outputs are in a different anatomical space, I think you would be fine just doing the additional brainstem segmentation (but if you want to be extra safe, you can rerun fmriprep inputting the output freesurfer folder).