AD Detection on in-house OCTA dataset (test)
85.32AccuracyPolar-Net
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| Polar-NetPrior Knowledge Matrix=true2023.11 | 85.32 | 0.8523 | 0.5817 | |
| Polar-NetFull Proposed Model=true2023.11 | 85.18 | 0.8484 | 0.5766 | |
| Polar-NetTransformation=false2023.11 | 83.88 | 0.8401 | 0.5279 | |
| Polar-NetPFEM=false2023.11 | 81.91 | 0.8315 | 0.5279 | |
| ResNet-34Model Version=34 layers2023.11 | 81.25 | 0.796 | 0.4909 | |
| MUCO-Net2023.11 | 79.68 | 0.7773 | 0.3985 | |
| EfficientNet-B3Model Version=B32023.11 | 79.42 | 0.7908 | 0.4177 | |
| ViT-BaseModel Version=Base2023.11 | 79.04 | 0.7726 | 0.3641 | |
| VAN-B6Model Version=B62023.11 | 77.07 | 0.6911 | 0.2939 | |
| HorNet-SGFArchitecture=Recursive Gated Convolution2023.11 | 76.02 | 0.5921 | 0.1738 | |
| SwinV2-TModel Version=Tiny2023.11 | 76.01 | 0.7528 | 0.3242 | |
| ConvNeXt-SModel Version=Small2023.11 | 75.62 | 0.5903 | 0.166 |