Metal artifact reduction in CT based on adaptive steering filter and nonlocal sinogram inpainting

The reduction of metal artifact reduction in CT has important clinical implications. This article adopted the adaptive steering filter to diminish noise content and smooth streak artifacts in the original CT image and utilize the means-shift and mutual information maximized segmentation (MIMS) to extract the metal component and artifact component respectively. After subtracting sonogram of metal artifact by the sonogram of original CT image, we complete the subtracted sinogram using the nonlocal means inpainting and reconstructed the corrected image by the filtered back-projection. And a final image from a combination of metal component and the corrected image are acquired. The results reveal that we effectively reduce the streaking artifacts in both phantom and real clinical image.