Stochastic image reconstruction for non-proliferation applications

The stochastic origin ensembles (SOE) method has been implemented for the reconstruction of neutron and gamma-ray images from a dual-particle imager (DPI). SOE image reconstruction is an iterative stochastic process that creates images using Markov-chain Monte-Carlo sampling with event origins from the reconstructed backprojection cones. This method has been shown to produce image quality comparable to that of the widely implemented maximum-likelihood expectation-maximization (MLEM) algorithm. SOE can be advantageous because the method is simple to implement and thorough definition of spatial efficiency or resolution is not required. The SOE algorithm also reaches a steady-state solution after iterating, as opposed to MLEM solutions, which can degrade with over-iteration. This paper shows photon and neutron images images reconstructed using the SOE algorithm for an experiment with multiple neutron and gamma-ray sources.