Line Artifact Removal Over Medical X-Ray Images via Joint Dictionaries Sparse Decomposition with Structured Incoherence

Line artifact is one of several artifacts that can affect the image quality of X-ray digital radiography. It appears as a sequence of a bright line followed by a darker one with variable intensity over the image. Extracting this artifact is challenging since it can remove some of the medical content and worsen the image quality. In this paper, we present a new line artifact suppression method based on a joint sparse dictionary model where the clinical content dictionary is learned with an incoherence penalty with respect to the artifact model. Tests on a large X-ray dataset show better estimation performance compared to the state-of-the-art methods.