Assessment of Beam-Hardening Correction Approaches for Ordered-subset Simultaneous Algebraic Reconstruction Technique

Studies have shown that X-ray computed tomography (XCT) measurements based on algebraic reconstruction technique (ART) algorithms require fewer projections than the traditional FDK algorithm, making ART a promising option for reducing acquisition time. However, because ART and FDK are based on different reconstruction processes, they may respond quite differently to errors in projection data. Among these errors, beam-hardening is particularly critical, as the use of polychromatic X-ray beams violates the Beer–Lambert law assumed in reconstruction. Moreover, widely used beam-hardening correction (BHC) approaches often rely on user-selected parameters, introducing additional uncertainties into the corrected projections. Although the BHC approaches have shown effectiveness combined with the FDK algorithm, it remains unclear whether they can support the XCT measurements using ART algorithms, especially when projections are reduced. This study investigates two practical BHC approaches on an aluminum prismatic part suffering from beam-hardening effects. The first approach selects an optimal preset coefficient for the BHC function from several candidates, while the second applies an automatic self-calibration procedure. The ordered-subset simultaneously algebraic reconstruction technique (OS-SART), a representative ART algorithm, is used for reconstruction, with the FDK algorithm combined with the first BHC approach serving as a reference representing the conventional strategy. The results provide guidance on selecting suitable BHC approaches for XCT measurements using ART algorithms, especially in applications where rapid acquisition is required.