Using previous Freesurfer output


What flags should I use if I want to specify the recon-all results I have from fmriprep, or just from using freesurfer itself? Just wondering if possible to avoid re-running everything!



By default fMRIprep will look for a folder called “freesurfer” (case-sensitive) in the output directory for pre-made derivatives. You can change this behavior by specifying --fs-subjects-dir.

Just saw the QSIprep tag, I’m not sure qsiprep uses FreeSurfer derivatives and it does not run recon-all by default, so you probably don’t need to worry about it.

Thanks Steven. That’s good to know. I thought that it’d need the FreeSurfer parcelations for WM and CSF segmentations? Or maybe when computing the connectivity matrixes during recon?

Out of curiosity, why implement the ability to run recon-all from within qsiprep if it won’t use it?



I think fMRIPrep uses the surface outputs to use bbregister registration methods. I don’t know why it may be suited better for one modality over another.

recon-all was left as an option in qsiprep for users who wanted to get a gm/wm segmentation from freesurfer. By default the segmentation is done with FAST, but either will work. The gm/wm segmentation is used for the anatomically-constrained tractography reconstruction workflows.

Would the choice to use Freesurfer instead of FAST for ACT reconstruction also influence the B0-T1 registration earlier on?
And is there a way to have the B0-T1 registration use a pre-existing preprocessed T1w images from a previously run fmriprep pipeline?

QSIPrep uses ants for everything. In testing, we found bbregister worked poorly in enough test cases that it was more robust to use ants instead.

This has been requested a lot and is at the top of our features-to-add list. Hopefully soon!