Table 1.
Comparison with state-of-the-art methods for pancreatic tumor segmentation on MSD dataset.
| Method | Dice | SEN | SPE |
|---|---|---|---|
| U-Net++ (38) | 0.4289 | 0.3966 | 0.9971 |
| Attention Unet (39) | 0.4317 | 0.5714 | 0.9977 |
| ResNet50 (40) | 0.4695 | 0.6056 | 0.9972 |
| Res_UNet (41) | 0.5264 | 0.7110 | 0.9960 |
| U-Net (14) | 0.4913 | 0.6846 | 0.9971 |
| Dense-UNet (42) | 0.5276 | 0.6133 | 0.9977 |
| nnUnet (43) | 0.5456 | 0.7268 | 0.9981 |
| C2FNAS (44) | 0.5636 | 0.6856 | 0.9982 |
| V-NAS (45) | 0.5730 | 0.6325 | 0.9942 |
| U-Shiftformer (47) | 0.5307 | 0.6833 | 0.9970 |
| DHT-Net (48) | 0.5618 | 0.6779 | 0.9975 |
| HyperSegNAS (46) | 0.5488 | 0.7123 | 0.9968 |
| IAG-Net (49) | 0.5762 | 0.7342 | 0.9969 |
| Ours | 0.5924 | 0.6824 | 0.9988 |
The bold value indicates the maximum value under the same indicator.