A Maximum-Likelihood Method for Quantitative 3D Source Reconstruction

Assumptions regarding source localization introduce uncertainties to holdup measurements that are often large and difficult to quantify. Gamma-ray imagers can angularly localize radioactive sources, constraining uncertainties from unknown source geometries. This benefit can be extended further by incorporating multiple viewing angles of the scene so that a full 3D distribution is measured. Iterative reconstruction provides a convenient way to seamlessly integrate imager data from multiple viewing angles. This paper presents an iterative technique based on Bayes’ theorem, maximum-likelihood expectation maximization. Challenges of the technique are presented along with methods to overcome or mitigate those challenges. Topics include background handling, data-driven convergence criteria, extended source reconstruction, and the importance of basis set restriction. Additionally, the maximum-likelihood technique has undergone testing in a laboratory setting to assess the expected total measurement uncertainty for a single imager measuring a point source.