Multi-Energy Sinogram Space Quantification for the IAEA PGET Detector

In this work, we further improve our previous solution that won the 2019 PGET tomographic reconstruction challenge. We propose an algorithm for gamma emissive tomography, to assess the absence of pins in spent fuel arrays, by applying joint emission (EM)-attenuation (AC) reconstruction algorithms translated from nuclear medicine applications. The IAEA PGET detector is used to independently assess the presence of used Uranium pins in spent fuel assemblies. It is based on a multi-energy detector pixel detector with a parallel hole collimator, that rotates around the detector. Due to the presence of large amounts of Uranium in the field of view, the biggest hurdle to perform correct quantification is Compton scattering. Joint EM/AC algorithms are not robust in general, and especially when normalization factors and scattering contribution are not exactly known. We overcome this issue by exploiting prior knowledge and by performing the EM and AC optimization in independent energy windows. The current version of the detector provides 4 energy windows, and we iteratively estimate the activity of each fuel pin in the highest energy bin, the least affected by attenuation, and we confirm their absence by looking at their attenuation in the lower energy bin, where the attenuation effect is higher. Scatter is approximated as the difference between the current sinogram estimate and the measured one, imposing constraints on the maximum admissible spatial frequency and on the scatter sinogram shape. In this work we show the mathematical principles and we test the algorithm on 1 datasets In this case, the proposed algorithm provides 100% accuracy for the identification of missing pins. The algorithm, as is, can be generalized to use as many energy windows as can be provided by improved detectors, to further improve robustness and quantitative accuracy.