Hi
> I have 3 subjects, each scanned twice, under two different conditions, 3D and 2D, where subjects passively viewed a 3D video clip and a 2D video clip in separate sessions without a model. If I want to compare the two conditions, would I first use melodic concat-ICA and then use dual regression to do this?
>
Yes, that's one option
> If so, for the melodic inputs, do I set up the 3D group and the 2D group inputs separately, or do I put them all together in the melodic gui setup?
>
Both is technically possible, though I'd use all data
> When running the dual regression command: dual_regression <group_IC_maps> <des_norm> <design.mat> <design.con> <n_perm> <output_directory> <input1> <input2> <input3>...1. How do I determine the number of permutations?
1/{numer of perms} is the lowest possible p-value you can assess, i.e. 1/200 = 0.005. I suggest using soemthing like 5000.
> 2. How do I set up the design.mat and design.con, if it was a model free design?
>
These are design matrices for the between-subject/session analysis (for which you have a model) _not_ the within session. In your case this is a paired t-test (3 subjects, scanned twice)
> I've test run dual regression just to see what outputs I'd get, but I do not understand what they mean or what to do with the outputs. Is there another step I have to do after running dual regression?
>
I can't comment on things like 'I've test run dual regression just to see what outputs I'd get' - maybe you have specific questions about specific output files?
hth
Christian
> Thanks,
> Amy
|