Yes, thanks Ed! The clusters are indeed highly significant.
Here's what I am trying to accomplish: I want to look at relatively small regions (such as the hypothalamus and other stuff in the midbrain) in the context of a whole brain analysis condition contrast that has a lot of activation. I want to avoid upping the cluster Z size--but increasing the p value so I can discern small regions that remain active at a stringent criteria. In other words, is there a way to run an analysis keeping the cluster size smallish--say 1.65--but increasing the p value dramatically so as to accomplish a correction for multiple comparisons that isn't solely based on the spatial correction? Does this make any sense. I can do this in FSLview, but that isn't legit since I have no idea how to actually analyze the data with those parameters. Another way to put this is are there any GUI ways to effect a stringent correction for multiple comparison that doesn't involve ramping up the cluster Z to values where I will lose my small ROIS? Any suggestions would be greatly appreciated. Thanks so very much!
|