Simulation for CZT Compton PET (Maximization of the efficiency for PET using Compton event)

Abstract Multiple interactions in positron emission tomography (PET) using scintillators are generally treated as noise events because each interacted position and energy of the multiple interactions cannot be obtained individually and the sequence of multiple scattering is not fully known. Therefore, the first interaction position, which is the crucial information for a PET image reconstruction, cannot be determined correctly. However, in the case of a pixelized semiconductor detector, such as CdZnTe, each specific position and energy information of multiple interactions can be obtained. Moreover, for the emission of two 511 keV radiations in PET, if one radiation deposits all the energy in one position (photoelectric effect) and the other radiation undergoes Compton scattering followed by the photoelectric effect, the sequence of Compton scattering followed by the photoelectric effect can be determined using the Compton scattering formula. Hence, the correct position of Compton scattering can be determined, and the Compton scattering effect, which is discarded in conventional PET systems can be recovered in the new system reported in this study. The PET system in this study, which was simulated using GATE 5.0 code, was composed of 20 mm×10 mm×10 mm CdZnTe detectors consisting of 1 mm×0.5 mm×2.5 mm pixels. The angular uncertainties caused by Doppler broadening, pixelization effect and energy broadening were estimated and compared. The pixelized effect was the main factor in increasing the angular uncertainty and was strongly dependent on the distance between the 1st and 2nd interaction positions. The effect of energy broadening to an angular resolution less than expected and that of Doppler broadening was minimal. The number of Compton events was double that of the photoelectric effect assuming full energy absorption. Therefore, the detection efficiency of this new PET system can be improved greatly because both the photoelectric effect and Compton scattering are utilized. On the other hand, since the angular resolution is strongly dependent on the pixelization effect, a finer pixel size and proper distance between the 1st and 2nd interaction position were needed to optimize the performance of CZT Compton PET. The total number of effective events detected by CZT Compton PET was 2.75 times higher than that by the same system using the photoelectric effect only.