Retinal OCT layer segmentation on Topcon dataset
1.37ILM ErrorCE-Net
Evaluation results
| Method | Links | |||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| CE-Netloss=Dice2019.03 | 1.37 | 2.02 | 2.08 | 1.8 | 2.47 | 1.48 | 1.1 | 1.26 | 1.48 | 1.74 | 1.68 | |
| U-Net2019.03 | 1.38 | 3.05 | 2.7 | 2.77 | 3.3 | 2.34 | 1.86 | 2 | 2.42 | 2.65 | 2.45 | |
| CE-Netloss=Cross Entropy2019.03 | 1.45 | 2.48 | 2.2 | 2.08 | 2.55 | 1.66 | 1.19 | 1.04 | 1.52 | 1.82 | 1.8 | |
| Topcon built-in2019.03 | 1.61 | 2.09 | 2.1 | 2.27 | — | 2.17 | 1.82 | — | 1.65 | 1.8 | — | |
| SRR2019.03 | 1.61 | 2.02 | 2.02 | 1.91 | — | 1.86 | 1.63 | — | 1.62 | 1.8 | — | |
| FCN2019.03 | 2.1 | 4.41 | 3.77 | 4.54 | 4.78 | 4.52 | 3.84 | 4.36 | 5.06 | 7.88 | 4.53 | |
| BackboneEncoder=Pretrained ResNet block2019.03 | 2.13 | 2.7 | 2.52 | 2.2 | 2.79 | 1.91 | 1.26 | 1.6 | 2.02 | 2.7 | 2.18 |