An efficient parallel architecture for 3D PET image reconstruction

Positron emission tomography (PET) is a functional imaging modality which provides information on in vivo physiological and metabolic functions in the human body. However, the long reconstruction time has prevented 3D PET from being practical. To make 3D PET feasible in a medical environment, in this paper, we propose an efficient parallel architecture for PET image reconstruction. The proposed parallel system is composed of four-connected processors with a novel interconnection network, called pairwise perfect shuffle (PPS). It may accomplish all types of data sharing activities involved in both filtered-backprojection and EM algorithms in O(logN) steps, which is optimal. Both proposed parallel architecture and reconstruction algorithms have been shown to be efficient on a transputer-based system.