Table 1.
Automatic segmentation results of different methods.
| Method | Dice (%) | JI (%) | Pre (%) | Recall (%) | ASD (voxel) | 95HD (voxel) |
|---|---|---|---|---|---|---|
| 3D U-Net (10) | 59.4 ± 4.8 | 42.4 ± 4.8 | 64.0 ± 6.1 | 55.9 ± 6.4 | 15.3 ± 14.2 | 31.4 ± 11.1 |
| nnU-Net (32) | 60.2 ± 3.5 | 43.2 ± 4.1 | 60.2 ± 6.6 | 61.9 ± 9.3 | 18.3 ± 19.1 | 35.7 ± 22.2 |
| DAF3D (14) | 60.8 ± 4.2 | 43.8 ± 4.2 | 64.1 ± 3.8 | 58.3 ± 7.0 | 14.7 ± 12.5 | 29.1 ± 10.6 |
| 3D FPN (12) | 59.3 ± 3.6 | 42.2 ± 3.7 | 64.9 ± 8.9 | 56.2 ± 8.8 | 17.2 ± 15.7 | 34.6 ± 15.2 |
| Ours | 62.6 ± 3.4 | 45.5 ± 3.6 | 67.1 ± 6.1 | 61.7 ± 5.4 | 14.2 ± 8.9 | 28.2 ± 9.9 |
Mean ± SD, with best results highlighted in bold.