Hi FSL experts,
I'm currently working on preprocessing some resting state fMRI data, and was hoping for some opinions on a question I have. Sorry in advance if it is inappropriate in anyways or quite basic, but here it goes.
So I've conducted the standard preprocessing steps, however my question pertains to removing potential noise artifacts that may occur in relation to motion/white matter/ventricular signals/ physio signals.
From my understanding, when using MELODIC ICA for denoising purposes, it is quite similar to if you were to model the typical sources of noise that I mentioned above through regressors. Using MELODIC ICA, I can visually inspect the components and can see which ones are related to these sources of noise such as the ventricles, motion, or physio related; I then can remove these from my data and continue on with my analysis.
I just wanted to ask if my thinking is correct in that using MELODIC ICA for denoising can provide the same results as modelling your nuisance regressors? From what I've been able to read, I don't recall ever seeing both methods used for denoising purposes. So far, I've been using MELODIC ICA for denoising, but I wasn't sure as to whether if it was okay to continue to use this, switch only to modelling my nuisance regressors, or perhaps conducting both.
Wouldn't surprise me if there wasn't a strict answer available, but any opinions would be great.
Thank you.
Paul
