Table 4 Comparison results of SSIM and PSNR with different landmarks input methods on DRIVE dataset
From: Retinal image synthesis from multiple-landmarks input with generative adversarial networks
| SSIM | PSNR | |
|---|---|---|
| Single vessel | 0.9417 | 23.469 |
| Fusion with one channel | 0.9449 | 22.069 |
| Fusion with multi-channel | 0.9498 | 23.722 |