Figure 2.

(a) Flowchart of conditional generative adversarial network (CGAN), and the architecture of the generator (b) and discriminator (c). The network consists of one generator and one discriminator with a conditional argument. The overall network’s performance is enhanced through each network acting bidirectionally. The artifacts in sparse projection are corrected by a network that maps images from a source domain (with artifact image) to the target domain (artifacts correction image) based on the conditional ideal image pair.