Improved Nonlocal Means for Low-Dose X-Ray CT Image

Abstract

Low-dose Computed Tomography (LDCT) can effectively lower the risk of the photon radiation during the CT examinations, however, the images reconstructed under the low dose protocol tend to be severely degraded by noise and streak artifacts. Therefore, how to enhance image quality as the normal dose scanning has attracted more and more attentions among recent several decades. This work aims to improve LDCT image quality through an improved nonlocal means (INLM). The proposed INLM method improves the original NLM method by calculating the weight map from a preprocessed one. CT images reconstructed under different doses from a Siemens CT with 16 detector rows are employed in experiments. Compared with the original NLM method, the proposed technique illustrates superior noise suppression in both simulated and real LDCT datum.