Hi everyone, 

To conduct surface-based FC analysis, the manual shows that we should enter the T1 or brainmask volumes in the subject-specific mri folder generated by Freesurfer, but it still uses the voxel-based functional images.

Here, I conducted surface-based preprocessing using fMRIPrep, and obtained L/R func.gii file for each task session. How can I use these files to perform surface-based FC analysis in conn?

Thanks in advance!

Hi everyone,

I'm wondering the same thing: How to import surface based functional data from fMRIPrep preprocessing, i.e. left and right hemisphere .gii files?

As I understood it you have to input T1 structural files from freesurfer preprocessing and functional files in subject space and run functional preprocessing using CONN's surface based pipeline. This makes preprocessing with fMRIPrep redundant - all you need is the freesurfer preprocessing of structural files.

Did I get it wrong?

Best wishes

Julian

Hi Julian,

I have not tried this yet, but at least in principle, it should be just a matter of using conn_surf_gii2nii to convert the lh.func.gii and rh.func.gii files into a single NIFTI format func.surf.nii file similar to the one that is produced by CONN's own surface-based preprocessing pipeline, so you could then simply import these files into CONN for surface-based analyses without any additional preprocessing required.

There may be some difficulties involved, as CONN uses freesurfer high-resolution format (with 163842 vertices and 327680 faces per hemisphere), and I am not sure if fMRIPrep uses the same convention. If in doubt, if you could please share your lh/rh.func.gii files I will be happy to take a look and see if I can propose a simple way to import that into CONN.

Best
Alfonso
Originally posted by Julian Dronse:

Thanks Alfonso for your quick response!

FMRIPREP outputs fsaverage5 space functional data, so not the same resolution as CONN.
I attached two hemisphere files as an example.
If there was a way to import these, that would be great!

Best wishes

Julian

please find the surfaces for left and right hemisphere attached

Hi Julian,

Would it be possible in your context to run fmriprep using the flag "--output-space fsaverage" instead of "--output-space fsaverage5" to create the files in the appropriate resolution? (fsaverage5 has only 10,242 vertices per hemisphere, while fsaverage has 163,842, so resampling-up from fsaverage5 to fsaverage is not ideal)

Best
Alfonso
Originally posted by Julian Dronse:

Hey Julian,

Wondering if you ever solved this? I am encountering the same difficulties right now.

Thanks!
Alex