A methodical assessment of interpolation techniques to metal artifact reduction in Computed Tomography

Metal artifact reduction (MAR) methods are continuously used for minimizing artifacts in Computed Tomography (CT/CAT scans) and although they are largely used, there is a lack of suggestions to evaluate such methods (especially the interpolation-based ones). In this paper, we propose an approach for assessment of a MAR method based on interpolation techniques. We have used four interpolation techniques, evaluating their performances in two different scenarios: an artificial image simulation for quantitative assessment, and one real image for qualitative assessment. We have adopted the Sinogram Completion method, widely used for metal artifact reduction. The linear interpolation had presented better results in comparison to other interpolation techniques in both scenarios. The MAR method (Sinogram Completion) associated to the linear interpolation technique presents the most efficient results, which justifies the current researchers' interest in this particular technique.