Provided are new methods for enhancing the selection of true (T)
annihilation events relative to the inclusion of true scattered (TS) and
random (R) annihilation events in PET tomographs and thereby improving
the sensitivity and/or resolution of PET scanners. The methods include
reconstruction of Compton scattering interactions in the .gamma.-ray
detectors for determining the angles of incidence of the .gamma.-rays
received at the detectors and may utilize .gamma.-ray polarization
effects and electron recoil data associated with positron annihilation
and Compton scattering. The use of the .gamma.-ray polarization effects
provides an improved ability for selecting data corresponding to T events
while simultaneously suppressing data corresponding to TS and R events
during PET applications.