This question has been giving me a headache for a while so I'll define the problem before asking the question.
fMRI is a measure of Blood Oxygenation Level Dependent signal. Higher the blood oxygenation, higher the signal. Of course, there is noise and drift associated with the signal as well. The entire brain is imaged 185 times in our experiment, of which the first 5 volumes (measurements) are dropped. Each voxel (3D pixel) that makes up the 3D brain has a time-series associated with it, (consisting of 185 measurements).
Functional connectivity is defined as the temporal correlation between spatially remote neurophysiological events in the brain. Basically, correlation between two regions, say dlpfc-r and amy-l is the correlation coefficient between the time series of those two regions.
In our experiment, six subjects undergo two fMRI runs (resting state) pre and post treatment.
I extracted the time-series of the two regions of interest (ROIs) and computed the correlation coefficient between them for each subject to yield the functional connectivity pre and post treatment. I used Fisher's r-to-z transformation (http://vassarstats.net/rdiff.html) to get the p-value for the within subject (between visit) comparison. I did that for each ROI-pair for each subject (pre and post) and Bonferroni corrected it for multiple comparisons.
I need to generalize this result. How do I do a group comparison of correlation coefficients that will tell me if there is significant difference between pre and post for each ROI-pair, and in which direction? Thank you.