Lesion Segmentation on NSCLC-Radiomics (5-fold cross-validation)
53.86Dice SimilarityResEncU-Net
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| ResEncU-NetArchitecture=ResEncU-Net, Training Paradigm=BackSplit, #Params=101.86M2025.11 | 53.86 | 125.9383 | 42.68 | |
| U-NetArchitecture=U-Net, Training Paradigm=BackSplit, #Params=30.70M2025.11 | 52.56 | 142.9498 | 41.36 | |
| SegResNetArchitecture=SegResNet, Training Paradigm=BackSplit, #Params=18.79M2025.11 | 51.31 | 141.5307 | 38.42 | |
| SegResNetArchitecture=SegResNet, Training Paradigm=Standard, #Params=18.79M2025.11 | 50.61 | 142.8023 | 39.21 | |
| ResEncU-NetArchitecture=ResEncU-Net, Training Paradigm=Standard, #Params=101.86M2025.11 | 50.2 | 145.0108 | 38.79 | |
| U-NetArchitecture=U-Net, Training Paradigm=Standard, #Params=30.70M2025.11 | 49.69 | 156.4725 | 38.03 |