Improving spatial resolution in semiconductor gamma-ray detectors using a list-mode EM algorithm for fluence estimation

We present a method for applying the expectation-maximization (EM) algorithm to semiconductor gamma-ray imaging detectors to reconstruct images from sets of electrode data. The EM algorithm iteratively seeks an image described by a distribution of fluence, or gamma rays per unit area, that has the greatest probability of producing the observed set of data. The algorithm acts without estimating interaction positions for individual gamma rays. When applying the algorithm to data from a Cd_0.8Zn_0.2Te detector with square pixel electrodes spaced 125 μm we get a spatial resolution of 40 μm for 60 keV gamma rays, compared with 104 μm resolution when individual interaction positions are estimated