How to use freesurfer 7.2 in fmriprep

1

Hi,

I want to know how to use freesurfer 7.2 in fmriprep? Or can I run skull-strip in fmriprep and insert the outputs into freesurfer 7.2 by myself (run -autorecon1 -noskullstrip, -autorecon2 and -autorecon3 in freesurfer 7.2) and how to do it? Because I find that skull-stripped brains of my subjects in fmriprep are much better than freesufer 7.2. Please give me some advices.
Many thanks!

Best,
Wenjun

2

The next release of fMRIPrep will use FreeSurfer 7.2. I think the simplest thing at the moment is to wait a couple days.

3

That’s great!!! Thank you.
Another queations: most of the subjects’ skull-stripped brain are good in fmriprep, but some are better in freesurfer. So I want to choose the better one of two skull-strip methods for each subjects. But I’m not sure if it is ok to use the outputs of freesurfer and the outputs of the freesurfer in fmriprep to do next vertex-wise and parcel-wise analysis in one project.

My concerns: I have checked recon-all.log in ./source/freesurfer of the fmriprep outputs and find that -autorecon2-volonly and -autorecon3 -nosphere -nosurfreg -nojacobian_white -noavgcurv -nocortparc -nopial -noparcstats -nocortparc2 -noparcstats2 -nocortparc3 -noparcstats3 -nopctsurfcon -nocortribbon -nobalabels skip some processing stages.

My fmriprep command:fmriprep-docker /datadir /outdir participant --participant-label sub-001 --anat-only --fs-no-reconall --fs-license-file /license_dir

My thought: I ran -autorecon2 and -autorecon3 on the basis of outputs in ./source/freesurfer. And the results in stats folder didn’t change. But, I’m not sure if there are any changes to the other files. So is it necessary for me to do so? After all, it will take a lot of time.

4

Sorry for the wrong fmriprep command in the last reply!
command: fmriprep-docker /data_dir /output_dir participant --participant-label ${subj} --fs-license-file /license --anat-only

5

We wrap ReconAll up in the surface_recon_wf workflow, which includes the autorecon_resume_wf subworkflow. The second one shows how we split up the work. We have verified that it performs the same jobs in the same order; it just works to avoid some issues with FreeSurfer trying to do some things that require both hemispheres too early. Any extra -no* flags apart from the ones in those docs mean that we detected that task had already been run and did not re-run it.

If you ran these, then you’re fine. fMRIPrep will see that it’s done and not try to rerun.

6

Does it mean that I can use the outputs of freesurfer and the outputs of the freesurfer in fmriprep to do next analysis in one project. Specifically, if I have ten subjects, No.1 to No.9 subjects get nice skull-stripping in fmriprep, No. 10 get nice skull-stripping in freesurfer. Then I run “fmriprep-docker /data_dir /output_dir participant --participant-label ${subj} --fs-license-file /license --anat-only” in fmriprep for No. 1-9 and run recon-all in freesurfer for No. 10 and use the outputs(although they may be from different apps) to do next vertex-wise and parcel-wise analysis.

7

I konw if I ran recon-all in freesurfer first and then put the subject folder in fMRIPrep, fMRIPrep will see that it’s done and not try to rerun. But I did run it in fMRIPrep first (“fmriprep-docker /data_dir /output_dir participant --participant-label ${subj} --fs-license-file /license --anat-only”), and then ran -autorecon2 and -autorecon3 on the basis of FMRIPrep’s outputs in freesurfer.

Anyway, I’m still a bit confused whether the above situation is feasible, and if not, what should be done about it.
(The reason I reran -autorecon2 and -autorecon3 on the basis of FMRIPrep’s outputs in freesurfer(for No.1-9 ) is that I want to make sure that I can use the outputs of No. 1-9 from FMRIPrep and outputs of No. 10 from freesurfer to do next analysis.)
I don’t know if I made my point clearly. It is a bit tedious.
Please give me some advices. Many thanks!