Annihilation photon GAN source model for PET Monte Carlo simulation.

Following previous works modeling sources of particles with GAN, we extend the proof of concept for generating back-to-back pairs of gammas with timing information, typically for Monte Carlo simulation of PET imaging. A conditional GAN is trained once from a low statistic simulation in a given attenuation phantom and allows generating various activity source distributions. A new parameterization that improves the training is also proposed. An ideal PET reconstruction algorithm is used to evaluate the quality of the GAN. The proposed method is evaluated on NEMA phantom and CT patient image showing good agreement with reference simulations. Once trained, the GAN generator can be used as input source for Monte Carlo simulations of PET imaging systems decreasing the computational time, with speedups up to 400 according to the configurations.&#xD.