You would have to feed data from each node as you did for the ROIs at the moment. I can imagine this to be quite slow, only you can judge if it is worth the effort. With Gang's tool, it should be less cumbersome.

Regarding multiple comparisons, you would need to do it. You should consider the number of voxels that went into the map, rather than the number of nodes. Also, search this message board for postings on surface-based multiple correction approach from Moo Chung and Donna Dierker (Hanlon at the time).
Search for 'surface multiple comparison' on this message board. But do not use the large 'Search' box on the top right, use the little 'search' link on the top of the message board threads.

cheers,
ziad