Convolutional Neural Network-based Reconstruction for Positronium Annihilation Localization

Abstract

A hermetic novel detector composed of 200 Bismuth germanium oxide crystal scintillators and 393 channel silicon photomultipliers has been developed for positronium (Ps) annihilation study. This compact 4π detector is capable of simultaneously detecting γ-ray decay in all directions, enabling not only the study of visible and invisible exotic decay processes but also tumor localization in positron emission tomography for small animals. In this study, we investigate the use of a convolutional neural network (CNN) for the localization of the Ps annihilation synonymous with tumor localization. The 2-γ decay systems of the Ps annihilation from the 22Na and 18F radioactive sources are simulated using GEANT4. The simulated data sets are preprocessed by applying energy cuts. The spatial error in the XY plane from CNN is compared to that from the classical centroiding, weighted k-means algorithm. The feasibility of the CNN-based Ps an-nihilation reconstruction with tumor localization is discussed.