Abstract
We investigate a method to improve the position decoding for thick crystal versions (i.e., /spl ges/8 mm) of our continuous miniature crystal element (cMiCE) PET detector by more accurately modeling the detector's light response function (LRF). The LRF for continuous detectors varies with the depth of interaction (DOI) of the detected photon. This variation of the LRF can result in a positioning error for two-dimensional positioning algorithms. The effect is greatest for photons interacting directly over a PMT channel or near the edge of the crystal. We explore a method to improve positioning performance by deriving two look up tables (LUTs), from the collected data, corresponding to the front and back regions of the crystal. The DETECT2000 simulation package was used to investigate the light response characteristics for a 48.8 mm by 48.8 mm by 10 (8) mm slab of LSO coupled to a 64 channel, flat panel PMT. Using DETECT2000 the light collected for each PMT channel versus interaction location in three dimensions is known. The data are then combined to produce the two dimensional light collection histograms that one would experimentally measure. Light collection histograms that have markedly non-Gaussian distributions (e.g., bimodal or have a high-end tail) are characterized as a combination of two Gaussian functions, where each Gaussian function corresponds to a DOI region of the crystal. The results indicate modest gains in positioning accuracy are achieved near the central region of the crystal. However, significant improvements in spatial resolution and positioning bias are achieved for the corner section of the detector.

This publication has 7 references indexed in Scilit: