Use the onavg template with FreeSurfer#

We have prepared the onavg template so that it can be easily used with FreeSurfer. These files can be put under FreeSurfer’s subjects directory, and they can be used as an reference subject during resampling. Here is a step-by-step guide on how to use it with FreeSurfer.

Part 1: Download and organize the template files#

Step 1: Enter FreeSurfer’s subjects directory#

If FreeSurfer has been properly configured in your system, you can find the directory with echo $SUBJECTS_DIR or echo $FREESURFER_HOME/subjects.

First, let’s enter FreeSurfer’s subjects directory using command line. This directory is where the files of the template “subjects” (e.g., fsaverage) are located. The path of the directory can be different in your system.

cd ~/lab/freesurfer/subjects

Step 2: Download the template files#

wget "https://www.dropbox.com/s/rvg6ui3dqjx7v8i/tpl-onavg_freesurfer.zip"

curl -O -L "https://www.dropbox.com/s/rvg6ui3dqjx7v8i/tpl-onavg_freesurfer.zip"

Alternatively, you can open the URL in a web browser, and click the “Download” button to download the zip file and save it to FreeSurfer’s subjects directory.

Step 4: Uncompress the zip file#

Unzip the zip file to get the content.

unzip tpl-onavg_freesurfer.zip

(Optional) Remove the zip file to save some disk space.

rm tpl-onavg_freesurfer.zip

After uncompressing the zip file, FreeSurfer’s subjects directory should look like this:

ls

bert                fsaverage3  fsaverage_sym  onavg-ico32  rh.EC_average
cvs_avg35           fsaverage4  lh.EC_average  onavg-ico64  sample-001.mgz
cvs_avg35_inMNI152  fsaverage5  onavg-ico128   onavg-ico8   sample-002.mgz
fsaverage           fsaverage6  onavg-ico16    README       V1_average

Resample data using the onavg template#

After uncompressing the onavg template, it can be used directly by FreeSurfer’s mri_vol2surf command using the --trgsubject parameter.

For example, the typical syntax that fMRIPrep uses to perform resampling is:

mri_vol2surf --hemi ${LR} --interp trilinear -o ${OUT_FN} \
  --srcsubject ${SUBJ} \
  --reg ${WF_DIR}/bold_surf_wf/itk2lta/out.lta \
  --projfrac-avg 0.000 1.000 0.200 --mov ${NII_FN} \
  --trgsubject fsaverage5

To use the onavg template, simply replace “fsaverage5” with “onavg-ico32”, which has the same resolution (10,242 vertices per hemisphere).

mri_vol2surf --hemi ${LR} --interp trilinear -o ${OUT_FN} \
  --srcsubject ${SUBJ} \
  --reg ${WF_DIR}/bold_surf_wf/itk2lta/out.lta \
  --projfrac-avg 0.000 1.000 0.200 --mov ${NII_FN} \
  --trgsubject onavg-ico32

Example Python script#

Here is a Python script that calls mri_vol2surf and resamples data to the onavg template space using the files from fMRIPrep’s work directory. The script is also available at GitHub: forrest_vol2surf.py

Resampling using FreeSurfer 7.2 is performed directly. Resampling using FreeSurfer 6 is performed using the singularity image of fMRIPrep 20.2.7, which was generated from the fMRIPrep Docker image.

Hide code cell content 
import os
from glob import glob
import subprocess
from joblib import Parallel, delayed


if __name__ == '__main__':
    dset, fmriprep_version = 'forrest', '20.2.7'
    bids_dir = os.path.realpath(os.path.expanduser(f'~/lab/BIDS/ds000113'))
    sids = ['01', '02', '03', '04', '05', '06', '09', '10', '14', '15', '16', '17', '18', '19', '20']

    jobs = []
    for sid in sids:
        sd = os.path.expanduser(f'~/lab/fmriprep_out_root/forrest_20.2.7/output_{sid}/freesurfer')
        onavg = os.path.expanduser('~/lab/freesurfer/subjects/onavg-ico32')
        link = os.path.join(sd, 'onavg-ico32')
        if not os.path.exists(link):
            os.symlink(onavg, link)

        wf_root = os.path.join(
            os.path.expanduser(f'~/lab/fmriprep_work_root/{dset}_{fmriprep_version}/work_{sid}'),
            f'fmriprep_wf', f'single_subject_{sid}_wf')

        raw_bolds = sorted(glob(f'{bids_dir}/sub-{sid}/ses-movie/func/*_bold.nii.gz')) + \
            sorted(glob(f'{bids_dir}/sub-{sid}/ses-localizer/func/*_bold.nii.gz'))

        for raw_bold in raw_bolds:
            label = os.path.basename(raw_bold).split(f'sub-{sid}_', 1)[1].rsplit('_bold.nii.gz', 1)[0]
            label2 = label.replace('-', '_')
            wf_dir = (f'{wf_root}/func_preproc_{label2}_wf')

            nii_fn = f'{wf_dir}/bold_t1_trans_wf/merge/vol0000_xform-00000_merged.nii'

            for space in ['onavg-ico32', 'fsavg-ico32']:
                target = 'fsaverage5' if space == 'fsavg-ico32' else space
                for lr in 'lr':
                    out_dir = os.path.join(
                        os.path.expanduser('~/lab/nb-data/forrest/20.2.7/resampled'),
                        space, f'{lr}-cerebrum', '2step_freesurfer7.2')
                    out_fn = os.path.join(out_dir, f'sub-{sid}_{label}.gii')
                    if os.path.exists(out_fn):
                        continue
                    os.makedirs(out_dir, exist_ok=True)

                    lta_fn = f'{wf_dir}/bold_surf_wf/itk2lta/out.lta'
                    if not os.path.exists(lta_fn):
                        os.makedirs(os.path.dirname(lta_fn), exist_ok=True)
                        import nitransforms as nt
                        lta = f'{wf_dir}/../anat_preproc_wf/surface_recon_wf/t1w2fsnative_xfm/out.lta'
                        nt.linear.load(lta, fmt='fs', reference=nii_fn).to_filename(
                            lta_fn, moving=f'{sd}/sub-{sid}/mri/T1.mgz', fmt='fs')

                    cmd = [
                        'mri_vol2surf',
                        # '--cortex',
                        '--hemi', f'{lr}h',
                        '--interp', 'trilinear',
                        '--o', out_fn,
                        '--srcsubject', f'sub-{sid}',
                        '--reg', lta_fn,
                        '--projfrac-avg', '0.000', '1.000', '0.200',
                        '--mov', nii_fn,
                        '--trgsubject', target,
                        '--sd', sd,
                    ]

                    jobs.append(delayed(subprocess.run)(cmd))

                for lr in 'lr':
                    out_dir = os.path.join(
                        os.path.expanduser('~/lab/nb-data/forrest/20.2.7/resampled'),
                        space, f'{lr}-cerebrum', '2step_freesurfer6')
                    out_fn = os.path.join(out_dir, f'sub-{sid}_{label}.gii')
                    if os.path.exists(out_fn):
                        continue
                    os.makedirs(out_dir, exist_ok=True)

                    lta_fn = f'{wf_dir}/bold_surf_wf/itk2lta/out.lta'

                    cmd = [
                        'FS_LICENSE=$HOME/FS_license.txt',
                        'mri_vol2surf',
                        # '--cortex',
                        '--hemi', f'{lr}h',
                        '--interp', 'trilinear',
                        '--o', out_fn,
                        '--srcsubject', f'sub-{sid}',
                        '--reg', lta_fn,
                        '--projfrac-avg', '0.000', '1.000', '0.200',
                        '--mov', nii_fn,
                        '--trgsubject', target,
                        '--sd', sd,
                    ]
                    cmd = ' '.join(cmd)
                    cmd = [
                        'singularity', 'exec', '-e',
                        '-B', '/dartfs:/dartfs',
                        '-B', '/scratch:/scratch',
                        '-B', '/dartfs-hpc:/dartfs-hpc',
                        '-H', os.path.realpath(os.path.expanduser(f'~/lab/singularity_home/fmriprep')),
                        os.path.expanduser('~/lab/fmriprep_20.2.7.sif'),
                        '/bin/bash', '-c', cmd,
                    ]
                    jobs.append(delayed(subprocess.run)(cmd))

    with Parallel(n_jobs=-1) as parallel:
        parallel(jobs)