I just started using
fmriprep and found it a great tool to preprocess resting state fMRI data. I am interested in using it to compute atlas-based connectivity matrices (conmats). Would the approach suggested below be adequate?
T1was output space (since conmats are computed in subject space):
fmriprep-docker /raw_data /preproc participant --nthreads 10 --output-space T1w --use-aroma --work-dir /tmp_data --write-graph --config /raw_data/nipype.cfg
Do non-agressive denoising, as suggested in
fsl_regfilt -i sub-<subject_label>_task-<task_id>_bold_space-T1w_preproc.nii.gz -f $(cat sub-<subject_label>_task-<task_id>_bold_AROMAnoiseICs.csv) -d sub-<subject_label>_task-<task_id>_bold_MELODICmix.tsv -o sub-<subject_label>_task-<task_id>_bold_space-<space>_AromaNonAggressiveDenoised.nii.gz
Highpass / lowpass temporal filters + scrubbing (any particular default suggestions?)
Mean of voxels in each ROI of the aparc+aseg atlas obtained in subject native space (
/preproc//freesurfer/sub-<subject_label>/mri/aparc+aseg.mgz). Is there any transform to apply to that file to match
sub-<subject_label>_task-<task_id>_bold_space-<space>_AromaNonAggressiveDenoised.nii.gzobtained in the previous step?
(Partial-) Correlation of the resulting mean ROI signals to obtain the connectivity matrix.
As an alternative, would you suggest to do the calculations in surface+volume space (GII format using the same aparc+aseg atlas)? In this case, how would it change the above workflow and associated commands?
Any help would be greatly appreciated.