Metal Artifact Reduction in Helical Cone-beam Computed Tomography

Metal artifacts are a major problem in computed tomography (CT). Most traditional metal artifact reduction (MAR) methods require a forward projection step which is time-consuming and costs unacceptable time in helical cone-beam CT (HCBCT). In this work, we propose an MAR method in the projection completion way without forward projection. Our MAR method locates the metal-affected data from only rays passing metal area segmented out from reconstructions, which tremendously saves computation compared with a complete forward projection. The method automatically takes care of multi-metal cases and can be used in fully 3D scan directly. Both numerical simulations and phantom experiments on a helical scan are done to evaluate the performance. FDK algorithm and the helical backprojection-filtration (BPF) algorithm are used for reconstruction. In the helical FDK reconstruction a linear interpolation is used for MAR, while in the helical BPF reconstruction a linear interpolation and a second-order polynomial interpolation are used for comparison. The experimental results show that the proposed MAR method can improve the image quality significantly and the processing time is ignorable compared to the step of reconstruction.