Med-CMR：融合视觉证据与临床逻辑的医学复杂多模态推理细粒度基准
Med-CMR: A Fine-Grained Benchmark Integrating Visual Evidence and Clinical Logic for Medical Complex Multimodal Reasoning
为医学多模态大模型提供按能力维度拆解的细粒度评测协议：把视觉理解拆成小目标检测、细节判别、空间理解，把推理拆成时序预测、因果推理、长尾泛化、多源整合，可直接用于诊断自研医学 VLM 的失效环节，尤其是长尾病例的视觉证据整合能力。
原文依据 long-tail generalization emerges as the dominant failure mode
摘要译文
由模型翻译 · 术语统一 · 请对照原文核对关键结论多模态大语言模型（MLLMs）已开始进入临床工作流，但其执行复杂医学推理的能力仍不明确。本文提出 Med-CMR，一个细粒度的医学复杂多模态推理基准。Med-CMR 与现有同类基准的区别体现在三个核心特征：1）系统化的能力分解，将医学多模态推理拆分为细粒度视觉理解与多步推理，从而支持有针对性的评测；2）具有挑战性的任务设计，视觉理解覆盖三个关键维度（小目标检测、细节判别、空间理解），推理覆盖四个临床相关场景（时序预测、因果推理、长尾泛化、多源整合）；3）广泛且高质量的数据覆盖，包含 20,653 个视觉问答（VQA）对，横跨 11 个器官系统与 12 种成像模态，并通过严格的两阶段（人类专家 + 模型辅助）审核验证，以确保临床真实性。我们使用 Med-CMR 评测了 18 个最优 MLLMs，结果显示 GPT-5 是表现最好的商用模型：多项选择题（MCQ）准确率 57.81，开放式评分 48.70，优于 Gemini 2.5 Pro（MCQ 准确率 49.87，开放式评分 45.98），并领先开源模型 Qwen3-VL-235B-A22B（MCQ 准确率 49.34，开放式评分 42.62）。然而，专门的医学 MLLMs 并不能稳定地优于强通用模型，且长尾泛化成为最主要的失效模式。因此，Med-CMR 为医学 MLLMs 的视觉-推理整合能力与罕见病例鲁棒性提供了一次压力测试，并为未来临床系统提供了严格的衡量标尺。
英文原文摘要 1693 字符
MLLMs are beginning to appear in clinical workflows, but their ability to perform complex medical reasoning remains unclear. We present Med-CMR, a fine-grained Medical Complex Multimodal Reasoning benchmark. Med-CMR distinguishes from existing counterparts by three core features: 1) Systematic capability decomposition, splitting medical multimodal reasoning into fine-grained visual understanding and multi-step reasoning to enable targeted evaluation; 2) Challenging task design, with visual understanding across three key dimensions (small-object detection, fine-detail discrimination, spatial understanding) and reasoning covering four clinically relevant scenarios (temporal prediction, causal reasoning, long-tail generalization, multi-source integration); 3) Broad, high-quality data coverage, comprising 20,653 Visual Question Answering (VQA) pairs spanning 11 organ systems and 12 imaging modalities, validated via a rigorous two-stage (human expert + model-assisted) review to ensure clinical authenticity. We evaluate 18 state-of-the-art MLLMs with Med-CMR, revealing GPT-5 as the top-performing commercial model: 57.81 accuracy on multiple-choice questions (MCQs) and a 48.70 open-ended score, outperforming Gemini 2.5 Pro (49.87 MCQ accuracy, 45.98 open-ended score) and leading open-source model Qwen3-VL-235B-A22B (49.34 MCQ accuracy, 42.62 open-ended score). However, specialized medical MLLMs do not reliably outperform strong general models, and long-tail generalization emerges as the dominant failure mode. Med-CMR thus provides a stress test for visual-reasoning integration and rare-case robustness in medical MLLMs, and a rigorous yardstick for future clinical systems.