Paper Overview
Field: Computer Vision (CV) Authors: Sicheng Yang, Hangjie Yuan, Wenjun Zhang Published: 2026-06-12 arXiv: 2606.14697
Summary
Building trustworthy medical multimodal large language models (MLLMs) is critical for reliable clinical decision support. Existing medical hallucination benchmarks mainly focus on data collection, but often ignore where hallucinations originate within the reasoning process. The authors find that hallucination sources vary across samples: errors may arise from visual misrecognition, incorrect medical knowledge recall, or flawed reasoning integration.
To enable source-level hallucination diagnosis, the authors introduce ClinHallu, a benchmark for stage-wise hallucination diagnosis in medical MLLM reasoning.
Key Contributions
- 7,031 validated instances, each augmented with a structured reasoning trace decomposed into three stages:
- Visual Recognition
- Knowledge Recall
- Reasoning Integration
- Stage-replacement interventions to measure how correcting a specific stage affects the final answer.
- Trajectory supervised fine-tuning shown to reduce stage-wise hallucinations.
Conclusion
ClinHallu provides a fine-grained hallucination testbed for diagnosing and mitigating reasoning failures in medical MLLMs. The benchmark is publicly available at https://github.com/alibaba-damo-academy/ClinHallu.
*Auto-collected on 2026-06-16.*