PARALLELPROJ -- An open-source framework for fast calculation of projections in tomography

In this article, we present a new open source framework, called parallelproj, for fast parallel calculation of projections in tomography using multiple CPUs or GPUs. This framework implements forward and back projection functions in sinogram and listmode using Joseph's method, which is also extended for time-of-flight PET projections. In a series of tests related to PET image reconstruction using data from a state-of-the-art clinical PET/CT system, we benchmark the performance of the projectors in non-TOF and TOF, sinogram and listmode using a multi CPU, hybrid CPU/GPU and pure GPU mode. We find that the GPU mode offers acceleration factors between 20 and 60 compared to the multi CPU mode and that OSEM listmode reconstruction of real world PET data sets is possible within a couple of seconds using a single state-of-the-art GPU.