Sparse-view muon computed tomography of an operating research reactor

Abstract

Directional radiography maps and projected computed tomography images of the ZED-2 research reactor facility at Canadian Nuclear Laboratories (CNL) have been reconstructed using a sparse, asymmetric, and time-restricted set of muon flux measurements. The accuracy of the radiography maps in representing the dominant infrastructure surrounding each detector position confirms MuPIC’s radiography capabilities for the first time in the field. Computed tomography images are reconstructed using variants of a least squares conjugate gradient technique, the Simultaneous Iterative Reconstruction Technique (SIRT), and the L1NORM technique, of which SIRT variants are found to be the most stable overall. Despite significant artifacts due to the sparsity of the data set, the dominant features of the reactor well are visible in the computed tomography projections—including reactor well walls, the graphite calandria, and fuel structure—further providing a first field validation of CNL’s computed tomography algorithm.