fMRIPrep: Does resampling in AROMA workflow lead to all-zero vertices in surface space?

1

Hi all,

fMRIPrep 1.2.4 outputs preprocessed (MNI152NLin2009cAsym_desc-preproc_bold) and AROMA-denoised (MNI152NLin2009cAsym_desc-smoothAROMAnonaggr_bold) BOLDs in MNI space. When projecting preprocessed BOLDs into surface space, everything looks fine. However, when projecting AROMA-denoised BOLDs into surface space, some non-midline vertices in some participants become all zeros.

This is the command I’m using:

		mri_vol2surf --hemi $hemi \
			--cortex \
			--mov $fmriprep_path/sub-${subj}_ses-001_task-grasp_run-001_space-MNI152NLin2009cAsym_desc-smoothAROMAnonaggr_bold.nii.gz \
			--regheader sub-${subj} \
			--out $results_path/sub-${subj}_ses-001_task-grasp_run-001_space-fsaverage5_desc-Denoised_hemi-$hemi.gii \
			--projfrac-avg 0.000 1.000 0.200 \
			--srcsubject sub-${subj} \
			--trgsubject fsaverage5 \
			--interp trilinear \
			--srchit $results_path/sub-${subj}_ses-001_task-grasp_run-001_space-fsaverage5_desc-srchit_hemi-$hemi.nii.gz

I noticed that the data dimensions between the preprocessed (78 x 93 x 78 x time) and AROMA-denoised data (91 x 109 x 91 x time) are not the same in volume space; might this obscure the surface projection for the AROMA-denoised data? Is there a setting in FSL so that my data is not being resampled in regfilt? What is the preferred way to project AROMA-denoised data into surface space?

Thanks for your help!

2

UPDATE: BOLDs are only being resampled when directly running fsl_regfilt but not when using fsl_regfilt via nipype. Possibly my quiet outdated version of fMRIPrep didn’t use the (current) nipype implementation back in the days. I will leave the post as is for future reference.