Opportunities and limitations of pure Cherenkov TOF PET scanners

• Main Authors: Razdevsekc, G., Dolenec, R., Pestotnik, R., Krizan, P., Rodriguez, D. Consuegra, Seljak, A., Studen, A., Orehar, M., Korpar, S.
• Format: Conference Proceeding
• Language: English
• Subjects: Image quality; Limiting; Logic gates; Microwave integrated circuits; Monte Carlo methods; Semiconductor detectors; Software
• ISSN: 2577-0829
• DOI: 10.1109/NSSMICRTSD49126.2023.10338809

To detect annihilation photons, PET conventionally employs scintillation light detection, but an interesting alternative involves detecting Cherenkov photons. Dense Cherenkov radiators are highly efficient at stopping gammas and can offer exceptional coincidence time resolution (CTR). However, due to the limited number of detected Cherenkov photons, pure Cherenkov detectors provide very little energy information, leading to higher scatter fractions. This work explores the performance of scanners based on pure Cherenkov detectors using Monte Carlo simulations. The simulations are performed on a super-computing network using GATE software, and CASToR software is used for (TOF-OSEM) image reconstruction. The investigated detectors are based on 20 mm long PbF_2 crystals wrapped with a reflector and coupled to a SiPM. The geometry of Cherenkov scanners is based on clinical scanners with a regular and long axial field of view (LAFOV) - Siemens Biograph Vision (Quadra). The performance of Cherenkov PET scanners is evaluated and compared to the reference scanners, following the NEMA standard and also imaging a highly anatomically detailed phantom (XCAT). Special attention is dedicated to the dark count rate (DCR) of the SiPMs, which could notably affect the image quality. The DCR of SiPMs is characterized at cryogenic temperatures, and the degree of cooling is estimated to make the impact of noise on image quality negligible. The results indicate that pure Cherenkov PET scanners can have a significant edge over scintillation-based scanners for imaging objects of diameters up to roughly 30 cm. This makes them especially suitable for brain and breast imaging, as well as preclinical imaging. Additionally, the results show that Cherenkov detectors are scalable. The low-cost Cherenkov detectors could therefore become very interesting for LAFOV scanners as their current high cost is limiting their dissemination in hospitals and research clinics.