In addition to Gang and Ziad's points, note that if
these bilateral ROIs are defined at the group level,
then you can have 3dGroupInCorr generate the
group results via the MASKAVE method in -batch
mode.
If the ROIs vary across subjects, that is not an
option.