RT Journal Article SR Electronic T1 A statistics-based positioning method based on Gaussian mixture model for a thick continuous slab detector JF Journal of Nuclear Medicine JO J Nucl Med FD Society of Nuclear Medicine SP 2048 OP 2048 VO 52 IS supplement 1 A1 SeungBin Bae A1 Kisung Lee A1 Changwoo Seo A1 Khiem Phung A1 Jinhun Joung YR 2011 UL http://jnm.snmjournals.org/content/52/supplement_1/2048.abstract AB 2048 Objectives Accurate position decoding of incidence photons on the monolithic scintillator block is one of the key issues to achieve high resolution for scintillation-based photon detectors. In this study, we developed a high precision position decoding method for a PET detector which consists of a thick slab scintillator coupled with a multi-channel PMT. Methods The DETECT2000 simulation package was used to investigate the light response characteristics for a 48.8 mm by 48.8 mm by 10 mm slab of LSO coupled to a 64 channel, flat panel PMT. The detector’s light response function for continuous detectors varies with the depth of interaction(DOI) of the detected photon so that the light histogram collected through all depths contains non-Gaussian distributions. We employed Gaussian mixture model(GMM) to parameterize the composite light response with multiple Gaussian mixtures regardless of DOI. Results The average spatial resolutions after positioning with four mixtures were 1.1 mm and 1.0 mm FWHM at the corner and center sections respectively. The bias in the position estimate was negligible for both central and corner sections of the crystal. Conclusions While the proposed positioning method showed modest gains in positioning accuracy near the central region of the crystal, it produces significant improvements in spatial resolution and positioning bias for the corner section of the detector Research Support This work was supported by the Basic Atomic Energy Research Institute (BAERI) (2010-0018616) and The Converging Research Center Program through the Ministry of Education, Science and Technology(2010K001193)Measured bias and spatial resolution of the decoded positions