Low-dose CT I mage D enoising M ethod B ased on Generative Adversarial Network

Abstract

In order to solve the problems of artifacts and noise in low-dose computed tomography ( CT ) images in clinical medical diagnosis, an improved image denoising algorithm under the architecture of generative adversarial network ( GAN ) is proposed. First, a noise model based on Style GAN2 is constructed to estimate the real noise distribution, and the noise information similar to the real noise distribution is generated by as the experimental noise data set. Then, a network model with encoder-decoder architecture as the core based on GAN idea is constructed, and the network model is trained with the generated noise data set until it reaches the optimal value. Finally, the noise and artifacts in low-dose CT images can be removed by inputting low-dose CT images into the denoising network. The experimental results show that the constructed network model based on GAN architecture can improve the utilization rate of noise feature information and the stability of network training, remove image noise and artifacts, and the reconstructed image has rich texture and realistic visual effect.