hi Shant
I think the problem here is that it is unlikely that the measure you want (which might be given by using 3dmean on the R-squared subbrik) would really be all that useful. it doesn't really make sense to me at least, because you will be averaging together a bunch of values only some of which are meaningful--for instance, if you are dealing with task-based functional data then the R-squared in the white matter voxels and ventricles is is really just noise.
perhaps it would be clearer if you gave an example of a paper where the measure you are asking for is used? or at least why you want this measure?
James
Edited 1 time(s). Last edit at 10/26/2012 07:19AM by jkeidel.