Many thanks Rick and Isaac for your comments. I'm amazed by how many things have already been implemented in AFNI.
As for the statistical analysis, do you think the one I proposed is sound? This relates to another confusing point in the paper: Although their wGBC and uGBC statistical models should be almost identifical, tables 3 and 4 (uGBC results) show T-statistics, although the thresholding should be done using the group ANOVA F-value (as discussed in "The wGBC statistical model").
Cheers,
John