Dear SPMers,
Sorry to post this message again, but I still have this question.
I am trying to perform a ROI analysis using a task protocol (passive viewing emotional stimuli) but I have some doubts about it.
I have two groups of subjects with 3 runs per subject performing the task in two different sessions. In a first level analysis I've modelled my two regressors (view negative pictures and view neutral pictures) plus movement parameters as no interest regressors, over the 3 runs (3 sessions specification). Then, for each subject I have calculated two simple contrasts: ViewNegative [1 0 zeros(1,6) 1 0 zeros(1,6) 1 0 zeros(1,6) 0] and ViewNeutral [0 1 zeros(1,6) 0 1 zeros(1,6) 0 1 zeros(1,6) 0].
After that, I have extracted beta values from my con images for the ViewNegative condition using a mask from bilateral amygdala (wfu pickatlas). Looking to the extracted values, I have found both negative and positive beta values, aproximately 60% of voxels are positive, depending on subject.
My question is if I am interested in evaluating the ViewNegative condition across group and sessions, should I restrict my analysis only to the positive beta values into the mask? How could I interpretate those negative beta values, considering that the amygdala is a region known to be activated following a negative stimulus presentation?
Reading the emails on this list, I was wandering if the right way to do this is to run second level analysis on each group, select voxels with p<0.05, select this cluster and extract beta values from individuals, and then compare between groups and sessions.
Hope anyone can help me to better understand these questions.
Thank you very much
|