Performance prediction of a high-performance whole-body human PET scanner by simulation

High-performance whole-body human PET scanners with excellent spatial resolution (< 2.5 mm) can potentially increase information content in the images for more accurate diagnosis. However, existing whole-body scanners have limited spatial resolution of ~3 mm or greater. This study proposes a high-performance PET scanner to address this need and predicted its performance by Monte Carlo simulation. The proposed PET scanner has 8 detector rings, and each ring consists of 32 dual-ended readout detectors. Each detector has two SiPM arrays coupled to both ends of a 32 × 32 array of 2.1 × 2.1 × 20 mm
³
LYSOs. The pitch size of the LYSO arrays is 2.2 mm. This scanner has a ring diameter of 780 mm and an axial length of 624.8 mm. The proposed scanner was simulated using GATE (v8.2), and the images were reconstructed using a TOF list-mode OSEM algorithm with image-space point spread function. Our simulation shows that radial spatial resolutions at the axial center with using DOI were 1.4, 1.5 and 1.7 mm for the radial offsets of 1, 10, and 20 cm, respectively. The corresponding radial resolution without using DOI information are 1.5, 3.1 and 6.2 mm. The sensitivity is 69 cps/kBq at the center of the scanner. The Derenzo phantom images show that 1.6-mm rods are visually resolved over the entire FOV given the DOI used, and better digital Hoffman brain phantom images were obtained using the DOI information compared to without using the DOI. The proposed design can be extended in the axial direction to form a total-body PET system.