A CNN-Based Hybrid Ring Artifact Reduction Algorithm for CT Images
Shaojie Chang, Xi Chen, Jiayu Duan, Xuanqin Mou
Abstract
Ring artifacts degrade the quality of reconstructed images in cone-beam computed tomography (CBCT). In this article, we propose a hybrid ring artifact reduction algorithm in computed tomography (CT) images based on a convolutional neural network (CNN), which fuses the information from the image domain and sinogram domain corrected images. The proposed method consists of three steps. First, the database for CNN training is established, which consists of artifact-free, ring artifact, and sinogram domain corrected images. Second, the original and sinogram domain corrected images are input to the trained CNN to generate an image with less artifacts. Finally, we use image mutual correlation to generate a hybrid corrected image by fusing the information from ring artifacts reduction in the sinogram domain and output by CNN. Both simulated and real experiments were performed to verify the proposed method. The experimental results show that the proposed method can suppress the ring artifacts effectively without the introduction of structure distortion.