Ring artifacts suppression for X-ray CT images by fusion of dual-domain images based on improved UNet

Abstract

Abstract The ring artifact is one of the typical artifacts in X-ray Computed Tomography (CT) images. The existence of ring artifacts will reduce the image quality, change the structure and details of the image, and affect the interpretation of image information. How to effectively suppress ring artifacts has always been an important research direction in the industrial and medical CT fields. In this research, three experiments of CT scanning were designed by using the microfocus cone-beam CT system, we take the real CT image sequences as datasets, and design customized loss functions according to the characteristics of the ring artifacts based on the structure of UNet, in addition, a model for suppressing ring artifacts is designed both in the slice domain and sinogram domain. Then, the outputs of the dual domain are fused using the Nonsubsampled Contourlet Transform algorithm. The network model is trained and tested using the real datasets, then Peak Signal-to-Noise Ratio (PSNR), Structural Similarity (SSIM), and mean squared error (MSE) are used as the evaluation indicators of image quality. Finally, the proposed method is compared with the typical algorithms of artifact suppression, and the experimental results show that the method proposed in this research can protect the structure information in CT images while suppressing the ring artifacts to the greatest extent, and the PSNR, SSIM, and MSE of the processed images can respectively reach 39.3 dB, 98.9 and 5.2×e− 4.