When contrasting different runs (i.e.- few complete runs are under one condition (or block) and few complete runs are under another condition)
Is it possible to apply activation-based analysis in such a design?
(Or only connectivity and other pattern-based analysis, which doesn’t depend on the average value of the specific scan)
Is there any information in the average activation value of a specific run, and if yes- how can it be properly analyzed?
what is the most recommend normalization option, for activation-based analysis?
Do you mean that you only have a time course per run and that you propose to compare the averge BOLD activity across runs ?
I’m not aware of any such analysis but this is not really recommended as the baseline value recorded with fMRI has no absolute scaling.
Normally, there should be some kind of on/off paradigm, the effect of which is comparable across runs ?
Thanks- this is indeed the problem-
That normalization relative to the control condition may eliminate the (simple) effect of some manipulation between the runs- the “control conditions” (control blocks) are not same across runs due to some manipulation performed between the runs, so the analysis will not allow to compare the activation effect of this manipulation and I was wondering if this means that only connectivity based (or any analysis with metrics invariant between runs)- will allow analysis of the effect of the effect beween the runs.
Yes, connectivity seems the only reasonable option to me.