Performance Characterization of an SiPM-based Time-of-Flight Canon PET/CT Scanner

Introduction: The use of silicon photomultipliers (SiPMs) as photosensors has enabled novel PET detector designs with improved physical performance characteristics. We measured the performance of an SiPM-based, time-of-flight (TOF) PET/CT scanner developed by Canon Medical Systems Corporation (Otawara, Japan). The scanner design is based on lutetium-yttrium oxyorthosilicate (LYSO) scintillator material, pixelated and coupled individually to SiPM photosensors. This study evaluated its important performance characteristics over clinically-relevant count rates according to the NEMA NU 2-2018 standard.

Methods: The new SiPM-based PET/CT scanner has a 780-mm bore diameter and 27-cm axial field of view (FOV). The PET detector consists of 48 circularly-arranged detector units, each consisting of an array of 12 x 60 pixels. Each pixel measures 4.2 mm x 4.2 mm x 20 mm and is coupled individually (i.e. “one-to-one”) to an SiPM photosensor, thus providing complete coverage of the outward side of the scintillator array. In this study, the important performance characteristics of the PET system were evaluated. The energy discrimination capability was measured by placing a 2.4 mCi ⁶⁸Ge line source at the center of the scanner FOV and collecting coincidence events. A histogram of acquired events’ energies was constructed and corrected for random events; the peak determined by fitting a parabola to the data within approximately 10% of the peak value; and the FWHM calculated as between the histogram values at half of the peak. The energy resolution was recorded as a percentage of the peak energy. System sensitivity was measured using ¹⁸F-FDG line sources inside five metal sleeves as according to the NEMA NU 2-2018 standard. The timing resolution, the noise equivalent count rate (NECR) and the scatter fraction were measured versus activity concentration according to NEMA NU 2-2018 across clinically relevant count rates. For comparison, timing resolution was also measured using a conventional method, i.e. a centered 2.4 mCi ⁶⁸Ge line source in air. Image quality was characterized by filling the NEMA image quality phantom with a sphere to background ratio of 4:1 and scan for 4 minutes. The images were generated using a list mode based TOF-OSEM reconstruction incorporating resolution recovery with 4 iterations, 12 subsets and 2-mm image voxels. No post-reconstruction filter was applied. Regions of interests were drawn according to NEMA NU 2-2018 standard.

Results: The measured energy resolution was 11.4% FWHM for the ⁶⁸Ge line source at scanner center. The measured timing resolution was 255 ps for the same ⁶⁸Ge line source. The measured sensitivity was 13.5 kcps/MBq, both when the ¹⁸F line source was at the scanner iso-center and 10 cm off center. The timing resolution ranges from 257 ps at an average effective radioactivity concentration of 5.3 kBq/mL to 262 ps at average effective radioactivity concentration of 9.6 kBq/mL, i.e. a singles rate equivalent to a typical whole-body patient scan. At 9.6 kBq/mL average effective radioactivity concentration, the scatter fraction is 0.36 and the NECR is 144.7 kcps. The contrast recovery coefficient ranges from 67.2% for the 10 mm diameter sphere to 93.1% for the 37 mm diameter sphere. The corresponding background variability ranges from 7.3% to 3.3% in the 10 mm to 37 mm background ROIs, respectively. Conclusions: The new SiPM-based PET/CT scanner from Canon demonstrates state of the art performance with 13.5 kcps/MBq sensitivity. Excellent timing resolution was measured across clinically relevant count rates.