Monte Carlo-based evaluation of inter-crystal scatter and penetration in the PET subsystem of three GE Discovery PET/CT scanners

While there is continuing demand for higher resolution in PET systems the technological improvements are still challenged by the presence of inter-crystal scatter (ICS) and inter-crystal penetration phenomena in PET detectors, which play an important role in deterioration of the spatial resolution. Both ICS and penetration have deteriorative impact on spatial resolution of PET scanners because they can lead to inaccurate incident crystal assignments. As such, an understanding of the quantitative behavior of ICS and penetration can be beneficial whether for design of a more optimized PET detection system or for more accurate modeling of ICS and penetration effects within the image reconstruction system matrix in order to enhance the quality of reconstructed images. In this work we analyzed the quantity of ICS and penetrated events in the form of coincidences, in contrast with the other studies that have assessed ICS and penetration in the form of single photons. This was performed in the PET subsystem of three GE whole-body PET/CT scanners: Discovery RX (DRX), Discovery ST (DST), and Discovery STE (DSTE). Furthermore, as a novel study, we discriminated between ICS vs. penetration events. In order to do this, we employed the GATE (Geant4 Application for Tomographic Emission) Monte Carlo (MC) toolkit for our simulations and used our previously validated GATE models of the scanners. Developing an algorithm, purely true coincidences were discriminated from ICS- and/or penetration-induced (ICSP) coincidences. ICSP coincidences were also categorized into three groups: group-1 consisted of coincidence event(s) only affected by penetration (one or both). Group-2 includes coincidences where one event is affected by ICS (possibly including penetration), while the other event is not affected by ICS (i.e. penetration or no mispositioning at all). Finally in group-3, both events are affected by ICS (possibly also including penetration). The results showed that the most magnificent quantitative variations of ICS-P occur along radial direction. In DRX, more than 55% of the true coincidences are mispositioned due to ICS and/or penetration when the source is located at the end of the transaxial field of view (FOV). This value for DST and DSTE is about 45%. Incidentally, the results revealed that the quantities of ICSP coincidences in the DST and DSTE are almost equal, while there is much smaller ICSP in the DRX.