Dear Allison,
>I have another PPI question. I have an fMRI study where subjects are looking
at images of faces with different emotional expressions. We are
investigating the interaction of the anterior cingulate and the amygdala
during different conditions. Many of the papers in the literature on PPI
show graphs where activity in one region is plotted vs. activity in another
region, under different conditions. (e.g. V1 vs. V5 in an attention
condition vs.an inattention condition). This is a relatively
straightforward thing to do if the design is a block design. I'm not sure
how to visually present what's going on in an event related design, though.
I've tried using the PPI utility to create deconvolved interaction vectors
of the VOI's and the individual conditions, and plotting those. For
example, I created a ppi vectors using the amygdala activity and the fearful
face condition, and then another one using the cingulate activity and the
fearful face condition, and plotted the PPI.ppi for the amygdala vs. the
PPI.ppi for the cingulate. Then, I can overlay the same thing for another
condition. However, I'm not sure that this is the correct way to go about
plotting cingulate activity vs. amygdala activity in the fearful face
>condition. I hope this makes sense. Does anyone have any advice?
That is a clever solution, which, in principle should work. The problem
in an event-related design is that the effects of the psychological
manipulation
(e.g. fear vs. neutral) may overlap in time. This means you cannot simply
regress amygdala activity on cingulate activity for fearful scans and then
for neutral scans. As you have proposed, you want to plot amygdala activity
against cingulate activity after removing any response component that
cannot be
explained by neutral faces and compare that ensuing regression with the same
plot having isolated the effect of fearful faces. These adjusted regional
activities can be computed using spm_regions (VOI) and the appropriate
contrast (or using the PPI deconvolution scheme with the appropriate
contrast).
I hope this helps - Karl
|