Detecting Clinical Hallucinations in LVLMs via Counterfactual Visual Grounding Uncertainty
Quick Answer
This paper shows that A new framework for detecting hallucinations in large vision-language models (LVLMs) enhances clinical image understanding by using visual evidence grounding.
Quick Take
A new framework for detecting hallucinations in large (LVLMs) enhances clinical image understanding by using visual evidence grounding. This method employs a counterfactual entity perturbation technique to improve detection accuracy, achieving better performance than recent baselines across various medical imaging modalities. The approach offers interpretable localization evidence and strong cross-model transferability.
Key Points
- Framework audits LVLM responses without modifying internal states.
- Uses a medical-domain-adapted Qwen-VL grounding verifier for entity localization.
- Introduces counterfactual perturbation to estimate visual evidence uncertainty.
- Achieves consistent improvements in hallucination detection performance.
- Code and dataset available at GitHub for further research.
Paper Resources
Article Content
From source RSS / original summaryarXiv:2606. 28520v1 Announce Type: new Abstract: Large (LVLMs) are increasingly used for clinical image understanding, yet they remain vulnerable to \emph{hallucinations}--producing textual findings or attributes not supported by the image. We present a vision-traceable hallucination detection framework that audits arbitrary LVLM responses via visual evidence grounding, requiring neither modification nor internal access to the hidden states of LVLMs.
Given an LVLM response, we extract visually verifiable entities and use a medical-domain-adapted Qwen-VL grounding verifier to localize each entity on the input image. To enhance the robustness of our detection method, we introduce a counterfactual entity perturbation method and estimate visual evidence uncertainty by contrasting factual and counterfactual grounding results.
Specifically, we compute an entity-level uncertainty score from the positive confidence, counterfactual confidence, and their grounding overlap for binary hallucination decision-making. Experiments on multiple medical imaging modalities and LVLM backbones demonstrate that our method consistently improves hallucination detection performance over recent baselines, while providing interpretable localization evidence and strong cross-model transferability. Code and dataset are available at https://github.
com/Agentic-CliniAI/CounterVHD.
Want this in your inbox every morning?
Daily brief at your local 8am — bilingual EN/中文, free.
More from arXiv cs.CV
See more →LLM-Guided ANN Index Optimization for Human-Object Interaction Retrieval
A phase-aware LLM agent optimizes human-object interaction retrieval, outperforming Optuna TPE by 33.3% and VDTuner by 34.2% on the HICO-DET benchmark. This method enhances throughput by 15.3x over UniIR and demonstrates strong transferability across vector database management systems.