Dear FSL experts,
I am currently analyzing data from a randomized control trial with 2 groups (treatment A/treatment B) who have been scanned pre- (t1) and post-treatment (t2) during resting state-fmri.
We are interested in the following questions:
Does group A significantly differ/improve with time compared to group B i.e. is the different at post-treatment present in group A but not group B (Group X Time interaction)
Do both groups improve over time irrespective of treatment? (Main effect of time)
Assumption: Both groups are similar at pre-treatment
Based on reading the list archives, I am following the design suggested in design_ekaterina.ods and running it in FEAT. However, I still have the following questions-
1. Is this design better/worse than one where the pre-treatment scan are used as covariates using a voxelwise EV and groups? How would such a design account for subject effects?
2. In a design where voxelwise EV would be used, should they be used a single covariate (one EV for both groups) or modeled separately for each group?
3. Is there an advantage to running these above designs in feat vs. randomise or PALM?
I would be grateful for any pointers!
Thanks,
Megha
|