Fig. 2From: Disentangled generative adversarial network for low-dose CTA basic architecture for proposed DNSGAN, where generator contains a dynamic filter network (DFN) and a residual dense network (RDN) to separately model denoising and structural restorationBack to article page