Low-Dose CT Image Denoising Using a Generative Adversarial Network Based on U-Net Network Structure

Low-dose Computed Tomography (CT) technology is widely used because it can greatly reduce the harm of scanning radiation to human body. However, due to the reduction of radiation, the projection data will be polluted and the reconstructed CT image will eventually have a lot of noise and artifacts. Therefore, how to improve the quality of CT images on the premise of reducing the radiation dose of CT scan has become a hot topic in the field of CT imaging. This article will combine the generative adversarial network (GAN) with U-net encoding and decoding structure applied in the low-dose CT (LDCT) image denoising study. Compared with other network structures, this network can extract image features more effectively and retain image details. At the same time design of stationary wavelet transform loss function, the realization of low dose CT effectively improve the quality of the images.