Hi,

If you have a look at this here https://neurostars.org/t/best-practices-... - I have asked the same question and came to the understanding that it seems safe to include WM and CSF calculated before ICA AROMA as additional confounders. That said, ICA AROMA should perform pretty well regardless of whether the additional physiological confounders are incorporated (e.g., Parkes et al., 2018 Neuroimage).