Table 1 SAFNet Achieves the Highest Mean DSC (%) for Most Individual Vertebra Segmentation
From: A spine segmentation method based on scene aware fusion network
Baseline | Fold_1 (%) | Fold_2 (%) | Fold_3 (%) | Fold_4 (%) | Fold_5 (%) |
|---|---|---|---|---|---|
3D UNet | 74.42 ± 5.22 | 72.45 ± 8.24 | 75.68 ± 7.42 | 74.22 ± 5.32 | 78.35 ± 3.22 |
CFNet | 78.95 ± 5.61 | 77.02 ± 8.14 | 78.31 ± 6.46 | 74.46 ± 5.67 | 80.13 ± 2.64 |
3D ResUNet | 78.95 ± 5.41 | 76.27 ± 9.07 | 79.04 ± 6.38 | 75.25 ± 5.76 | 80.07 ± 4.49 |
3D DeepLabV3 | 79.19 ± 6.52 | 78.62 ± 8.48 | 82.77 ± 3.46 | 72.59 ± 5.88 | 80.83 ± 3.48 |
SAFNet | 79.46 ± 4.63 | 78.82 ± 7.97 | 81.32 ± 3.45 | 80.56 ± 5.47 | 81.45 ± 3.47 |