Fully 3D GPU PET reconstruction

Abstract

Fully 3D iterative tomographic image reconstruction is computationally very demanding. Graphics Processing Unit(GPU) has be enproposed formany years as potential accelerators incomplex scientific problems, but it has not be enuseduntil there cent advances in the programmability of GPU sthat the best available reconstruction code shave started tobe implemented toberunon GPUs. This work presents a GPU based fully 3D PET iterative reconstruction sofware. This new code may reconstruct sinogram data from several commercially available PET scanners. The most important and time consuming part soft he code,the forward and backward projection operations, arebasedonan accurate model of the scanner obtained with the MonteCarlo code PeneloPET and they have been massively parallelize don the GPU.For the PET scanners considered, the GPU based code is more than 70 times faster than asimilar code running on a single core of a fast CPU, obtainingin both cases the same images. The code has been designed to be easily adapted to reconstruct sinograms from any other PET scanner, including scanner prototypes.