Adaptive nonlocal means-regularized iterative image reconstruction for sparse-view CT

Low-dose X-ray computed tomography (CT) imaging is desirable for various clinical applications due to the growing concerns about excessive radiation exposure to the patients. One strategy to achieve low-dose CT imaging is to lower the number of projection views per rotation during data acquisition. However, the resulting image by the conventional filtered back-projection method may suffer from view-aliasing artifacts due to insufficient angular sampling. In this work, we propose a nonlocal means (NLM)-regularized iterative reconstruction scheme for low-dose CT from sparse-view acquisitions. In order to improve the quality of reconstructed images, we further introduce spatial adaptivity to the NLM-based regularization by considering the local characteristics of images. The resulting approach is termed as adaptive NLM-regularized iterative image reconstruction. Experimental results demonstrated the feasibility of the presented reconstruction scheme for sparse-view CT and the superiority of incorporating the spatial adaptivity.