From Maneesh: to find the center of the RF, don't do center of mass,
and don't remove a constant and then find center.  Instead, threshold
(perhaps at half peak or the median) and *then* find the center.
