EVALUATION OF ITERATIVE INTERPOLATION METHODS FOR METAL ARTEFACTS REDUCTION OF PELVIC COMPUTED TOMOGRAPHY IMAGES: A PHANTOM STUDY

Metal artefacts obscure the organs as shown on a computed tomography (CT) image. The effect of the artefact can be reduced by using sinogram completion techniques. The aim of this study is to evaluate interpolation methods - linear interpolation (LI), total variation in painting (TV), high order total variation inpainting (HTV) for reducing the effect of the artefact. In this study, the sinogram completion technique applied on the simulated pelvic CT image of a hip prosthesis. Algorithms consist of four steps: 1) segment the metal region on the image, 2) transform the image into a sinogram and find the boundary of the metal sinogram, 3) estimate the new value of the missing data using three different methods 4) reconstruct the modified sinogram and add back the metal region into the reconstructed image. The percentages of the difference of intensity value between the artefact image and the non-artefact image of three techniques were computed. The percentages of artefact reduction were evaluated to compare the effectiveness of algorithms. The HTV method can provide the lowest difference of intensity value amongst three interpolation algorithms. The artefact reduction of the HTV method reduced 89.43%. The HTV technique could be used for solving the metal artefact problem effectively on CT images.