Hi,

I just discovered 3dABoverlap this past weekend and it was perfect timing. It was just a day or two before I would've started writing a bash script to try to use 3dcalc to obtain the same automask overlap stats as a cross-check of inter-session coregistration. So thank you! :)

In our case, however, we already have the automasks, and as we have other uses for them, the handling within 3dABoverlap doesn't save us a step either. Since the unique contribution of 3dABoverlap is the overlap stats part (not the automasking), it would be great if it didn't assume _it_ had to do the automasking. Please consider adding an option to indicate that A and B are themselves automasks, in which case the 3dAutomask-calling part can just be skipped. If you want, you could require that no resampling be necessary, but I'd encourage you to just resample the automasks. (Of course it would be better to resample the EPIs before automasking, but just mention that in the help--a user should be free to do it anyway if they want, IMO. :-p What if the EPIs aren't available for some reason?)

As the first two parts of 3dABoverlap I presume are just calls to 3dresample and 3dAutomask (and therefore not "unique contributions" of 3dABoverlap to the suite's functionality, so to speak), maybe the automasking preparatory step (with resampling as needed beforehand) should even be opt-in (e.g. "-automask") instead of opt-out.... (That question is really secondary to me, though. Just having the choice at all is most important.)

What do you think?

Thanks,
Terry