Hi Igor,

I'm not sure if I can be of much help here. You are faced with the difficult interaction between diagnosis and scanner. Besides the differences in scanner there might also be differences in the way the participants were handled/scanned between the different sites.

That said, the Biswal PNAS 2010 paper (and others since), have indicated that you can pool data from different centers. Despite marked differences between sites, we were still able to discern reliable resting state findings. However, in our opinion it is indeed imperative that you account for site differences. see also this message in the forums: http://www.nitrc.org/forum/message.php?msg_id=6050

Best,
Maarten