Quantitative imaging and automated fuel pin identification for passive gamma emission tomography

Compliance of member States to the Treaty on Non-Proliferation Nuclear Weapons is monitored through nuclear safeguards. The Passive Gamma Emission Tomography system a novel instrument developed by International Atomic Energy Agency (IAEA) for verification spent fuel stored in water pools. Advanced image reconstruction techniques are crucial obtaining high-quality cross-sectional images spent-fuel bundle allow inspectors IAEA monitor material and promptly identify its diversion. In this work, we have software suite accurately reconstruct cross sectional image, automatically present rods, estimate their activity. Unique challenges posed measurement fuel, due high activity self-attenuation. We implemented linear forward model detector responses rods inside PGET. performed solving regularized inverse problem using fast-iterative shrinkage-thresholding algorithm. also traditional filtered back projection method comparison applied both methods simulated mockup assemblies. Higher resolution fewer artifacts were obtained with inverse-problem approach, mean-square-error reduced 50%, structural-similarity improved 200%. then used convolutional neural network type extract pin locations from images; estimated levels finally being compared ground truth. proposed computational pins, an associated uncertainty approximately 5%.