A selective kernel-based cycle-consistent generative adversarial network for unpaired low-dose CT denoising

Abstract

ABSTRACT Low-dose computed tomography (LDCT) denoising is an indispensable procedure in medical imaging field, which not only improve the image quality, but can mitigate the potential hazard to patients caused by routine-doses. Despite the improvement in performance of Cycle-consistent adversarial network (CycleGAN) due to the well-paired CT images shortage, there is still needs for further reducing image noise while retaining detailed features. Inspired by residual encoder-decoder convolutional neural network (RED-CNN) and U-Net, we proposed a novel unsupervised model using CycleGAN for low-dose CT imaging, which injects two-sided network into Selective Kernel Networks (SK-NET) to adaptively select features, and uses patchGAN discriminator to generate CT images with more detailed maintenance, aided by added perceptual loss. Based on patch-based training, the experimental results demonstrated that the proposed SKFCycleGAN outperforms competing methods in both clinical dataset and Mayo dataset. The main advantages of our method lie in noise suppression and edge preservation.