An adaptive regularized iterative FBP algorithm with high sharpness for irradiated fuel assembly reconstruction from few projections in FNCT

Fast Neutron Computed Tomography (FNCT) can be applied in identifying induced structural anomalies of Irradiated Fuel Assemblies (IFA). This is because fast neutrons have high penetrability, and FNCT has the capability to avoid the interference of sample radioactivity and reconstruct internal information of sample. Few projections and a high sharpness are required for FNCT. In this study, an Adaptive Regularized Iterative Filtered Back-Projection algorithm (ARIFBP) was proposed to reduce the influence of few projections and improve sharpness by integrating the Filtered Back-Projection (FBP) operator into the Simultaneous Iterative Reconstruction Technique (SIRT), introducing Total Variation (TV) regularization, Median Root Prior algorithm (MRP) regularization and an adaptive parameter method. IFA and ammonal IFA from 120 simulated projections were reconstructed to evaluate the performance of ARIFBP algorithm. Their results have illustrated that the ARIFBP algorithm can better suppress the influence of few projections, obtain higher sharpness and more accurately identify anomalies of IFA than other three algorithms i.e. FBP, SIRT and SIRT-TV. Furthermore, the improved sharpness and fewer projections can be achieved by employing the projection difference data between IFA and abnormal IFA. (C) 2020 Elsevier Ltd. All rights reserved.