Medical report generation on IU-Xray (BLEU-2/BERTScore/RadGraph/F1)
28.4BLEU-2RadHiera
Evaluation Results
| Method | Links | |||||||
|---|---|---|---|---|---|---|---|---|
| RadHieraModel=RadHiera2025.11 | 28.4 | 60.7 | 62.6 | 31.3 | 2.325 | 51.9 | 91.5 | |
| R2GenGPTModel=R2GenGPT2025.11 | 27.1 | 55.8 | 60.1 | 26.9 | 1.751 | 48.3 | 85.2 | |
| CheXpertModel=CheXpert2025.11 | 26.4 | 51.6 | 60.8 | 25.7 | 1.619 | 46.9 | 82.8 | |
| MedGemma-4BModel=MedGemma-4B2025.11 | 25.9 | 57.1 | 60.3 | 26.8 | 1.821 | 50.2 | 89.1 | |
| MedVersaModel=MedVersa2025.11 | 25.1 | 57.9 | 61.1 | 27.4 | 1.852 | 49.4 | 89.8 | |
| RadFMModel=RadFM2025.11 | 22.8 | 49.1 | 58.9 | 25.4 | 1.32 | 42.1 | 84.1 |