First Pre-Clinical Study of Total-Body Dynamic PET Imaging using the mini-EXPLORER Scanner

Objectives: Non-human primate (NHP) PET imaging is of importance in translational molecular imaging research for the development of diagnostic and treatment methods for human diseases. Existing clinical whole-body PET scanners offer limited sensitivity and temporal resolution for NHP imaging. Our team has built a mini-EXPLORER PET scanner, dedicated to NHP imaging with an axial FOV of 45 cm, capable of covering the entire body of a rhesus monkey. It provides 5-fold increase in sensitivity compared with a clinical whole-body PET and offers a platform to explore the benefits of total-body imaging. Here we present our quantitative image reconstruction technique and the first dynamic total-body PET imaging using the mini-EXPLORER. Methods: The mini-EXPLORER was constructed using 192 detector modules of a Siemens mCT PET scanner. The system consisted of 8 axial rings, each formed by 24 detectors with an inner diameter of 43.4 cm. Each detector consisted of a 13x13 array of 4x4x20 mm3 LSO crystals. The system TOF resolution is 609 ps. We developed a 3D TOF list-mode OS-EM reconstruction with accurate resolution model and quantitative correction. Component-based normalization factors were estimated from a scan of an annulus source of 31-cm diameter. Attenuation factors were obtained using a separate CT scan aligned to the emission image. Scatters were estimated using the single scatter simulation algorithm and randoms were estimated using a model-based method from the delayed random sinogram in each time frame. To evaluate its capability for mid-size animal imaging, we conducted the first pre-clinical 18F-FDG PET study using a male rat, 26 cm in length (excluding tail) and weighing 431 grams. 1.14 mCi was injected. A total of twelve billion prompt events were acquired over a 2-hour dynamic scan. To demonstrate the improved image quality offered by the high sensitivity, dynamic PET data were reconstructed into 1-second, 10-second, 1-minute, and 10-minute frames. Furthermore, the list-mode data were down-sampled to generate datasets mimicking dynamic PET scans with 1/2, 1/5, 1/10 and 1/20 of the original injected dose, respectively. Regions of interest (ROIs) were drawn for major organs and the tissue time activity curves were fitted to a two-tissue-compartment model. The input function was extracted from the left ventricle for the first 40 seconds and the vena cava for the remaining 7,160 seconds. Results: Reconstructed dynamic PET images showed good image quality