The use of generalized moments for data reduction in maximum likelihood positioning for gamma cameras

The ML-GM (maximum-likelihood generalized-moment) method is shown to be worthwhile as a positioning technique, as there is a bijection between the space of the moments and the scintillation coordinates space. It allows efficient computation to be done, as the required dimensionality of the problem is kept at the minimum number required to take into account the 3-D nature of the scintillation process. Moreover, the generalized moments are more stable than individual photomultiplier-tube response. The computation can be made even faster by use of a precomputed look-up table of the GM vs. the coordinate (X, Y, Z, thus sparing the need to compute the moments online. The depth-of-interaction estimate makes the energy evaluation better than in Anger positioning. Any corrections required by the nonhomogeneous nature of the crystal do not interfere with the positioning method, but are done painlessly as an after-positioning computation. >