I am using fmriprep to preprocess my rsfMRI data and I am curious if appending ‘CSF’ and ‘WM’ columns from the *confounds.tsv (as well as perhaps other columns, e.g., ‘GlobalSignal’) to the MELODIC mixing matrix is an acceptable approach for simultaneously regressing out “noise” independent components and physiological signals. I have some Nipype code for doing this ( github.com/sjburwell/fmriprep_denoising ), but the process goes as follows:
- run fmriprep with the AROMA option turned ‘on’
- append columns in the *confounds.tsv I wish to remove (i.e., WM, CSF) to the *MELODICmix.tsv, save as e.g., *MELODICmixPLUSPhysio.tsv
- append the indices of those columns to the *AROMAnoiseICs.csv, save as *AROMAnoiseICsPLUSPhysio.tsv
- run fsl_regfilt, with the “design” file being *MELODICmixPLUSPhysio.tsv and the “filter columns” being *AROMAnoiseICsPLUSPhysio.tsv.
This approach should theoretically get around having to re-extract the WM, CSF, etc. from the *variant-smoothAROMAnonaggr_preproc.nii.gz file if I want to use ICA-AROMA in combination with physio confounds ( github.com/poldracklab/fmriprep/issues/817 ). The one concern I have, however, is whether there is an issue with having potentially “competing” physiological signals in both the “signal” and “noise” components of the fsl_regfilt equation. That is, there may be “WM” and “CSF” components in the MELODIC mixing matrix that aren’t flagged by ICA-AROMA (and thus kept as “signal”), which are competing for variance with the “WM” and “CSF” signals from the *confounds.tsv (which I explicitly define above as “noise”) in fsl_regfilt.
Any thoughts or prior experiences with this problem would be greatly appreciated!