During the last decade, Photon-Counting Computed Tomography (PCCT) has been quickly developing and it is expected to revolutionize the X-ray imaging field. Rapid development has also been ongoing in the scintillator and photosensor fields: fast inorganic scintillators have become available on the market and the silicon photomultiplier (SiPM) technology has made substantial progress. Therefore, SiPM-based scintillator detectors may soon be successfully applied in challenging applications like PCCT. In this work, a simulation framework is developed to evaluate detectors composed of scintillators coupled to SiPMs. The idea is to combine Monte Carlo simulations and a library to generate SiPM signals. Focus is given to six fast inorganic scintillators currently on the market (Ce:LYSO, Ce:LuAP, Pr:LuAG, Ce:LuAG, Ce:GGAG (ceramic), Ce:GAGG (single crystal)) and different TiO2-epoxy mixtures. The main properties of scintillating and reflecting materials to be defined in the simulation database are characterised experimentally. Using a virtual model created within this framework, an analysis of the performance in PCCT of a SiPM-based scintillator detector composed of the considered materials is performed. Simulated pulses are processed to report on the count-rate capability. Among the studied crystals, Ce:LYSO would enable sustaining the highest rate of interacting X-ray (2.5 Mcps/pixel with 30% of pile-up). Ce:GAGG could also handle a rate ~ 1 Mcps/pixel with identical pile-up conditions. Other materials show a slow decay time in their scintillation kinetics, implicating a < 1 Mcps/pixel count rate. A qualitative evaluation of the energy binning efficiency is also accomplished, by defining this parameter as a quality metric for multiclass classification. Scintillators with high light yield and good energy resolution (Ce:LYSO, Ce:GAGG and Ce:GGAG) present the best energy binning performance, as expected. The dependence of this aspect is explored as a function of the pulse processing method used, the crystals size and the pile-up probability. Resulting trends respect predictions, even though for a more quantitative analysis a more in-depth study is required.
Read full abstract