A pragmatic approach to metal artifact reduction in CT: merging of metal artifact reduced images

Abstract

The purpose of this study was to improve metal artifact reduction (MAR) in X-ray computed tomography (CT) by the combination of two artifact reduction methods. The presented method constitutes an image-based weighted superposition of images processed with two known methods for MAR: linear interpolation of reprojected metal traces (LI) and multi-dimensional adaptive filtering of the raw data (MAF). Two weighting concepts were realized that take into account mean distances of image points from metal objects or additional directional components. Artifact reduction on patient data from the jaw and the hip region shows that although the application of only one of the MAR algorithms can already improve image quality, these methods have specific drawbacks. While MAF does not correct corrupted CT values, LI often introduces secondary artifacts. The corrective impact of the merging algorithm is almost always superior to the application of only one of the methods. The results obtained with directional weighting are equal to or in many cases better than those of the distance weighting scheme. Merging combines the advantages of two fundamentally different approaches to artifact reduction and can improve the quality of images that are affected by metal artifacts.