医学视觉模型常错误选择否定答案，导致临床判断矛盾。
CXR-ContraBench: Benchmarking Negated-Option Attraction in Medical VLMs
- 构建胸部X光诊断基准，检测模型对否定选项的错误偏好。
- 主流模型在存在性问题上选否定项超62%，准确率仅30%左右。
- 提出无需重训的修复方法，使准确率提升至95%以上。
当胸部X光显示实变但问题询问哪个发现存在时，医疗视觉语言模型可能回答“无实变”。这不仅是错误选择，更是一种极危险的极性反转，与图像事实直接冲突。本文将此现象称为否定选项吸引，即模型被否定选项误导，即使视觉证据和问题指向相反。为此，我们提出CXR-ContraBench（胸部X光矛盾基准），涵盖内部ReXVQA子集及外部OpenI、CheXpert协议。基准聚焦存在性问题：若模型在可见病灶时仍选“无X”，则构成主要临床风险；而不存在性问题用于检验模型是否机械复制否定语句。在CheXpert协议下，该失败普遍存在且顽固。在严格直接存在性探测中，MedGemma和Qwen2.5-VL准确率分别仅为31.49%和30.21%；在匹配的135,754条记录训练集协议下，两者在存在性问题上选择否定选项比例均超62%。链式思维提示可缓解部分极性反转，但无法根除，反而可能加剧不存在性矛盾。最后，提出的QCCV-Neg（否定条件一致性验证器）在不重训情况下确定性修复极性混淆子集，使两模型在直接存在性探测中的准确率分别提升至96.60%和95.32%。结果表明，标准准确率可能掩盖具有临床意义的推理期极性错误。源代码与基准构建脚本见https://github.com/fangzr/cxr-contrabench-code。
原文摘要 · Abstract (English)
When a chest X-ray shows consolidation but the question asks which finding is present, a medical vision-language model may answer "No consolidation." This is more than an incorrect choice: it is a polarity reversal that emits a clinical statement contradicting the image. We study this failure as negated-option attraction, where a model is drawn to a negated answer option even when it conflicts with both the visual evidence and the question. We introduce CXR-ContraBench (Chest X-Ray Contradiction Benchmark), a diagnostic benchmark spanning internal ReXVQA slices and external OpenI and CheXpert protocols. The benchmark centers on present-finding questions, where selecting "No X" despite visible X creates the main clinical risk, and uses absent-finding questions as secondary tests of whether models copy negated wording. Across CheXpert protocols, the failure is substantial and persistent. On a strict direct presence probe, MedGemma and Qwen2.5-VL reach only 31.49% and 30.21% accuracy, respectively; on a matched 135,754-record CheXpert training-split protocol, both models select negated options on over 62% of presence questions. Chain-of-thought prompting reduces some presence-side reversals but does not eliminate them and can amplify absence-side contradictions. Finally, QCCV-Neg (Question-Conditioned Consistency Verifier for Negation) deterministically repairs the measured polarity-confused subset without retraining, raising MedGemma and Qwen2.5-VL to 96.60% and 95.32% accuracy on the direct presence probe. These results show that standard accuracy can hide a clinically meaningful inference-time polarity failure. Source code and benchmark construction scripts are available at https://github.com/fangzr/cxr-contrabench-code.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。