Clinical Report Generation on MIMIC-CXR (Accuracy, Localization, Professionalism, Clinical Admissibility)
6.04AccuracyEviAgent
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| EviAgent2026.03 | 6.04 | 6.32 | 8.7 | 6.61 | |
| Lingshu-7BParameters=7B2026.03 | 5.88 | 5.87 | 8.66 | 6.37 | |
| Gemini-2.5-Flash2026.03 | 5.74 | 6.19 | 9.06 | 6.48 | |
| MedGemma-4B-ITParameters=4B2026.03 | 5.44 | 5.61 | 8.16 | 5.97 | |
| GPT-5.12026.03 | 4.91 | 4.98 | 8.67 | 5.62 | |
| Qwen3-VL-8BParameters=8B2026.03 | 3.94 | 4.66 | 7.51 | 4.75 | |
| Claude 4.5 Sonnet2026.03 | 3.51 | 3.79 | 8.59 | 4.4 | |
| InternVL3-8BParameters=8B2026.03 | 3.07 | 4.71 | 7.55 | 4.05 | |
| InternVL2.5-8BParameters=8B2026.03 | 2.41 | 3.44 | 7.33 | 3.41 | |
| Qwen2.5-VL-7BParameters=7B2026.03 | 2.3 | 3.49 | 7.3 | 3.43 | |
| HuatuoGPT-V-7BParameters=7B2026.03 | 2.2 | 5.07 | 7.33 | 3.17 | |
| LLaVA-Med-7BParameters=7B2026.03 | 1.74 | 2.88 | 4.28 | 2.23 | |
| BiMediX2-8BParameters=8B2026.03 | 1.41 | 2.76 | 3.84 | 1.86 |