Medical Report Generation on MIMIC-CXR & OPEN-I Weighted Average benchmark leaderboard · SOTA2 Research

Medical Report Generation on MIMIC-CXR & OPEN-I Weighted Average

29.07BLEU-1DeepMedix-R1

Evaluation Results

Method	Links
DeepMedix-R12025.09		29.07	19.12	14.09	10.96	28.07	28.73	22.19	28.09	48.62	42.75	51.07	29.34
LLaVA-Rad2025.09		23.73	14.2	9.74	7.11	23.44	24.32	17.59	27.06	45.09	40.54	48.96	25.62
DeepMedix2025.09		22.84	16.41	13.13	11.36	22.72	26.06	19.79	20.58	40.19	31.54	40.44	24.1
HuatuoGPT-Vision2025.09		15.22	7.54	4.37	2.85	19.22	15.15	9.05	18.99	30.63	31.04	36.21	17.3
MedGemma2025.09		14.78	8.13	5.3	3.72	21.84	17.1	15.5	27.68	44.88	40.01	47.05	22.36
CheXagent2025.09		11.33	8.46	7.23	6.52	12.95	14.62	13.65	17.97	31.71	27.02	32.65	16.74
DeepSeek-VL-7B2025.09		9.39	4.04	2.34	1.62	13.52	13.03	4.99	12.68	25.44	10.88	15.37	10.3
Qwen2.5-VL-32B2025.09		7.01	3.38	1.91	1.2	17.06	11.89	7.48	19.24	29.38	28.43	30.67	14.33
Qwen2.5-VL-7B2025.09		6.42	3.83	2.95	2.51	7.52	9.72	5.03	12.38	20.82	15.81	17.26	9.48
Qwen2.5-VL-7B2025.09		6.25	3.05	1.75	1.11	15.74	11.58	8.23	15.61	24.38	21.81	24.2	12.16