Dear All,
I am trying to calculate the percent signal change (PSC) invoked by a particular experimental condition across multiple sessions. Specifically, I have 4 conditions of interest A1-A4 and I used a rapid ER design with 4 runs/sessions in total. Hence my contrast for Beta estimate pertaining to, say A1, would be repmat([1 0 0 0], 1,4). Now, my objective is to calculate the (local) PSCs pertaining to A1-A4 for each subject and then perform a group analysis on these extracted values.
My questions are as follows:
1) W.Penny suggested that (local)PSC = Beta(i)/Beta(Const)*100. However, P. Mazaika and J. Mumford indicated that in ER designs one must take into the height of an isolated event. I am wondering whether Penny's formula can be adoted in my case (as I am using a rapid ER design) ?
2) I wonder in case of multiple sessions calculation fo PSC in the above way is problematic because if there are multiple runs of data, it might be vulnerable to cross-run variation in baseline when done after individual subject analysis. Neither Dr.Penny nor Dr.Mumford has talked about this issue in their posts.
I wonder whether some sort of extra special measures should be taken in this context?
Many thanks and best regards,
Ce
