FastPET reconstruction in whole-body 82-Rubidium molecular imaging

Deep learning-based PET image reconstruction has gained increasing attention due to its fast reconstruction times and low-count imaging capabilities. Fast reconstruction is particularly valuable for dynamic PET imaging, where multiple time frames are needed to capture tracer kinetics. Long axial field-of-view PET scanners enable total-body quantitative blood-flow imaging, allowing measurements across multiple organs of interest beyond the myocardium. This study provides the first systematic evaluation of FastPET, which uses multi-angular histo- images and attenuation correction factors as inputs to a 3D UNET to reconstruct PET images in approximately 20 seconds, in whole-body 82-Rubidium blood flow imaging.