Lesion Detection on AbdomenAtlas 3.0
86.7Precision for Pancreatic LesionsnnU-Net
Evaluation Results
| Method | Links | |||||||||
|---|---|---|---|---|---|---|---|---|---|---|
| nnU-NetEvaluation Approach=Segmentation-Based Detection (RadGPT-style), Segmentation Configuration=Seg(F+L)2026.02 | 86.7 | 35.5 | 50.3 | 49.1 | 74.8 | 59.3 | 49.2 | 48.1 | 48.6 | |
| nnU-NetEvaluation Approach=Segmentation-Based Detection (RadGPT-style), Segmentation Configuration=Seg(C+L)2026.02 | 69.1 | 50.9 | 58.6 | 34.2 | 90.8 | 49.7 | 32.3 | 87.8 | 47.2 | |
| U-VLMEvaluation Approach=End-to-End Report Generation, Segmentation Configuration=Seg(C+L)2026.02 | 61 | 58.2 | 59.5 | 54.8 | 79.4 | 64.8 | 58.1 | 68.7 | 62.9 | |
| U-VLMEvaluation Approach=End-to-End Report Generation, Segmentation Configuration=Seg(F+L)2026.02 | 51.2 | 39.1 | 44.3 | 59.8 | 70.2 | 64.6 | 45.8 | 75.6 | 57.1 | |
| M3DEvaluation Approach=End-to-End Report Generation, Zero-shot=true2026.02 | 11.8 | 3.6 | 5.6 | 23.5 | 9.2 | 13.2 | 16.5 | 10.7 | 13 | |
| RadFMEvaluation Approach=End-to-End Report Generation, Zero-shot=true2026.02 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 |