Hi everyone,
I'm trying to model a Group x Time x HRF interaction with 3dMSS to see how HRFs in response to stimuli of certain groups change over time.
Some background:
I have data with three groups who underwent task-fMRI at two time points one year apart. Two groups have a disease at the first time point; one group recovers (recovered) and one does not (persistent). The third group is healthy controls (healthy). The task is ingestion of a milkshake or a tasteless control while undergoing fMRI.
For each subject and session, I have 9 bricks from 3dDeconvolve containing the estimated HRF response to the milkshake minus the HRF response to tasteless control with 9 beta-values per subject. These beta-values are from TENT functions centered at each TR. Each brick is called /path/to/subject/contrast_t{index}, which is calculated by subtracting milkshake and tasteless betas at each index.
I am able to use 3dMSS to analyze pairwise group differences (e.g. between every combination of healthy recovered and persistent) in HRF contrast at each time point separately (baseline and followup), but I'm wondering if I could use 3dMSS to model how the HRF changes over time across groups.
Right now I'm modeling them as separate groups and grouping by time point (e.g. with a term in the model like s(TENT, k=9, by=timepoint), but this doesn't directly compare across groups. If one group shows changes above some threshold in a certain region but another doesn't, it could just be because of the threshold we chose (although maybe the highlight don't hide approach would help with this?)
Is there a way to do this with 3dMSS? I thought about doing a sort of "double" contrast, where I take \text{contrast}_{t_2} - \text{contrast}_{t_1} and running this through 3dMSS, but I don't know if there is a better way to do it or if this is even valid.
Sorry for the long message, and thank you for making this software and these forums.
Caleb