Figure 1 | Scientific Reports

Figure 1

From: Prediction of anti-vascular endothelial growth factor agent-specific treatment outcomes in neovascular age-related macular degeneration using a generative adversarial network

Figure 1

Overview of the proposed post-therapeutic optical coherence tomography image synthesis model. Models are trained with a training set of each anti-vascular endothelial growth factor agent. For ranibizumab, the model consists of two attention-guided generators, G-ranibizumab and F-ranibizumab. For aflibercept, the model consists of G-aflibercept and F-aflibercept. Each generator aims to detect the discriminative pixel of the image. After that, the input image, content mask, and attention mask are mixed to manufacture the synthesized therapeutic image or reconstructed pre-therapeutic image. Moreover, the models are constrained by cycle-consistency loss.

Back to article page