Artifact reduction in dual-energy CT reconstruction for security applications

In this paper we present a structure-preserving dual-energy (SPDE) CT reconstruction method for security, which provides enhanced estimates of material-dependent basis coefficient images. In security applications many different materials may be scanned in various degrees of clutter and metal objects are common. Image noise and metal artifacts are often severe and can lower the accuracy of material identification. The proposed approach aims at mitigating metal artifacts and improving material parameter accuracy. The basis coefficient images are generated jointly as the solution of a single modelbased optimization problem. An auxiliary variable corresponding to a mutual boundary-field is estimated as well and applied to the basis coefficient images to reduce object splitting and material variability. In addition, metal aware data weighting is included to reduce streaks. We evaluate the performance of the method both visually and quantitatively using simulated dual-energy data. We demonstrate a significant reduction in noise and metal artifacts compared to a baseline filtered back projection (FBP) method.