Accurate computed tomography (CT) numbers derived from dual-energy CT (DECT) have numerous applications, including optimising the precision of radiation treatment planning. The dual-energy iterative reconstruction algorithm (DIRA) utilizes material decomposition in the iterative loop to generate monoenergetic images free of beam-hardening artifacts. In simulations, the reconstructed CT numbers closely matched reference tabulated values. This study evaluates the feasibility of applying DIRA to experimental data. Transitioning from simulations required several adaptations: (i) removal of the patient table, (ii) modeling of the bow-tie filter, and (iii) application of an inverse water beam-hardening correction. Axial scans of a cylindrical polymethyl methacrylate phantom containing four rod inserts of distinct materials were acquired using a clinical Siemens SOMATOM Force scanner. By the sixth iteration, DIRA yielded CT numbers that closely matched tabulated values. The algorithm also demonstrated strong performance on an anthropomorphic head phantom (Computerized Imaging Reference Systems, Inc. (CIRS) model 731-HN) with known material composition.
Magnusson et al. (Fri,) studied this question.