Least Squares Estimate Algorithm Based on Particle Swarm Optimization for Position Reconstruction of Scintillation Crystal

Reconstruction algorithm is a key technology for improving the performance of gamma cameras. This paper presents a least squares estimate (LSE) algorithm based on particle swarm optimization (PSO) to estimate the position of a scintillation event detected in a continuous crystal by a position sensitive photomultiplier tube (PSPMT). The theoretical framework of this algorithm consists of two parts: an enhanced objective function with a coefficient established by LSE via fitting the charge signals with the Scrimger-Backer formula; PSO method to search the global optimum solution of the objective function. The constant coefficient eta = Q/I introduced to objective function is demonstrated to effectively improve the position linearity response and expand the useful field of view (UFOV) up to almost the dimension of the PSPMT for both wide and narrow light distribution. More powerful global searching ability and highier computational efficiency are shown in PSO-based LSE algorithm in comparison with the gradient-descent-based LSE (GD-based LSE) algorithm. Furthermore, the PSO-based LSE algorithm overcomes the compression effect produced by trunction of light distribution in CoG algorithm and exhibits an almost idea position linearity response as well as 96 similar to 99% ratios of FOV/crystal size for different light distributions.