Hi,
Following up on this thread as I am attempting to run ICA/Dual Regression on a rsFC dataset that has between/within-subject factors to model. Briefly, I have two groups (Trauma, No Trauma); each participant received two scans (Drug vs. Placebo).
I looked on the FSL practicals (and other threads), but am confused about the ordering of operations for running Group ICA and Dual Regression--specifically for ultimately using randomise to test between/within-group differences (Mixed 2x2 ANOVA). As of now, I have done preprocessing, ICA-AROMA, and nuisance regression (as outlined in Pruim et al., 2015). Based on this thread and the FSL GLM page, it seems I need to subtract scans (i.e., Drug-Placebo, Placebo-Drug) and then conduct a two-sample t-test to examine group differences. Let me know if this is incorrect.
Question: At what point should I conduct subtraction?
For Group-ICA/Melodic, do I need to concatenate all participants (Trauma & No Trauma)/all scans (Drug & Placebo)? Or should I run them separately (i.e., run Melodic on Drug session, all participants concatenated; run Melodic on Placebo session, all participants concatenated)?
Apologies on any misunderstanding on my end, and thanks for your input!
Best,
Erin
|