Noise removal and motion correction task-based fMRI

, , , ,
1

Hi all,

I’m looking for input on motion/noise correction for task-based fMRI with a movement-related paradigm.

Participants perform movements that are related to the task, so motion is partly of interest and partly nuisance. I’m comparing ICA-AROMA vs aCompCor (fMRIPrep outputs) and am unsure which is more appropriate here.

Concerns:

  • ICA-AROMA may remove task-related signal if components correlate with movement.

  • aCompCor may be too conservative and leave residual motion artifacts.

Questions:

  • For movement-related tasks, do you prefer ICA-AROMA, aCompCor, or another approach?

  • Do you combine these with motion parameters (e.g., 24P/36P) or keep models minimal?

  • Any experience with AROMA removing task-relevant variance?

  • Additionally, what do you use as a framewise displacement threshold for scrubbing in a movement related task?

Happy to provide more details if helpful.

Thanks!