Pulmonary nodule segmentation on Multi-center external (test)
88.07Dice Coefficient (%)Hi-Seg
Evaluation results
| Method | Links | ||
|---|---|---|---|
| Hi-SegAnnotator group=junior medical student2026.06 | 88.07 | 0.682 | |
| Hi-SegAnnotator group=senior radiologist2026.06 | 87.48 | 0.579 | |
| Hi-SegAnnotator group=overall, five human annotators2026.06 | 87 | — | |
| Hi-SegAnnotator group=non-medical annotators2026.06 | 86.49 | 0.942 | |
| MedSAMPrompt type=bounding box prompt2026.06 | 81.5 | 0.001 | |
| MedSAM2Prompt type=overall, prompts #1-42026.06 | 80.9 | 0.001 | |
| HQ-SAMPrompt type=overall, prompts #1-42026.06 | 78.11 | 0.001 | |
| SAMPrompt type=overall, prompts #1-42026.06 | 73.85 | 0.001 | |
| nnU-NetModel type=Deep learning model2026.06 | 69.05 | 0.001 | |
| U-NetModel type=Deep learning model2026.06 | 66.52 | 0.001 | |
| U-Net++Model type=Deep learning model2026.06 | 65.25 | 0.001 | |
| Attention U-NetModel type=Deep learning model2026.06 | 63.25 | 0.001 | |
| TransU-NetModel type=Deep learning model2026.06 | 56.89 | 0.001 |