Table 8 Semi-supervised segmentation performance (Dice score, %) on the LIDC-IDRI dataset under different labeled data proportions
Method | 5% | 10% | 25% | 100% (Full) |
|---|---|---|---|---|
Voxel Baseline (U-Net) | 58.1 | 65.2 | 74.9 | 85.0 |
Mean Teacher | 61.8 | 67.5 | 77.9 | 86.0 |
FixMatch | 62.5 | 68.8 | 77.5 | 86.3 |
ST++ | 63.0 | 69.1 | 78.2 | 86.5 |
CoreFormer (Ours) | 68.5 | 74.1 | 81.5 | 88.8 |
