SHOVIR benchmark uses region-level occlusion of chest X-rays to show that top-scoring radiology report generators often rely on shortcuts rather than on the visible evidence.
hub
Detecting and Evaluating Medical Hallucinations in Large Vision Language Models
15 Pith papers cite this work, alongside 6 external citations. Polarity classification is still indexing.
15
Pith papers citing it
6
external citations · Pith
abstract
Large Vision Language Models (LVLMs) are increasingly integral to healthcare applications, including medical visual question answering and imaging report generation. While these models inherit the robust capabilities of foundational Large Language Models (LLMs), they also inherit susceptibility to hallucinations-a significant concern in high-stakes medical contexts where the margin for error is minimal. However, currently, there are no dedicated methods or benchmarks for hallucination detection and evaluation in the medical field. To bridge this gap, we introduce Med-HallMark, the first benchmark specifically designed for hallucination detection and evaluation within the medical multimodal domain. This benchmark provides multi-tasking hallucination support, multifaceted hallucination data, and hierarchical hallucination categorization. Furthermore, we propose the MediHall Score, a new medical evaluative metric designed to assess LVLMs' hallucinations through a hierarchical scoring system that considers the severity and type of hallucination, thereby enabling a granular assessment of potential clinical impacts. We also present MediHallDetector, a novel Medical LVLM engineered for precise hallucination detection, which employs multitask training for hallucination detection. Through extensive experimental evaluations, we establish baselines for popular LVLMs using our benchmark. The findings indicate that MediHall Score provides a more nuanced understanding of hallucination impacts compared to traditional metrics and demonstrate the enhanced performance of MediHallDetector. We hope this work can significantly improve the reliability of LVLMs in medical applications. All resources of this work have been released at https://github.com/ydk122024/Med-HallMark.
hub tools
citation-role summary
background 1
citation-polarity summary
years
2026 15roles
background 1polarities
background 1representative citing papers
MM-Snowball benchmark diagnoses hallucination snowballing in multi-turn MLLM dialogues; CAVR mitigates it via dual visual rectification at representation and logit levels.
HalluCXR benchmark shows 61.9-82.3% hallucination rates across VLMs on MIMIC-CXR images, identifies patterns such as length-based risk and over-fabrication of common findings, and demonstrates ensemble mitigation that cuts fabrication by up to 84.8%.
Mind-ParaWorld creates parallel worlds with atomic facts to evaluate search agents on future scenarios, showing they synthesize evidence well but struggle with collection, coverage, sufficiency judgment, and stopping decisions.
A counterfactual visual grounding uncertainty method detects hallucinations in LVLMs on medical images, improving over baselines with interpretable evidence and cross-model transfer.
Strong medical-model task scores do not guarantee process stability: omission, contradiction, and delay mainly break contradiction detection, diagnosis updating, and hallucination self-correction while final evidence grounding can look fine.
Reweighting training emphasis toward image-negative tokens and filtering hallucinated data reduces object hallucination in LVLMs across three model variants.
MedVIGIL provides a 300-case evaluation suite with 2556 probes that measures silent failures in medical VLMs under broken evidence, showing the best model at 69.2 on the composite score versus a human radiologist at 83.3.
CoEV is a plug-and-play bidirectional verification method that maps text statements to visual evidence regions, assigns them to a four-quadrant factuality-grounding map, and uses this to detect and correct hallucinations in medical VLMs without retraining.
VIHD detects hallucinations in medical MLLMs via visual dependency probing, token masking intervention, and calibrated semantic entropy on VQA benchmarks.
MPD reduces hallucinations in LVLMs by 23.4% while retaining 97.4% of general capability through semantic disentanglement and selective parameter updates.
MESA reduces hallucinations in LVLMs via controlled selective latent intervention that preserves the original token distribution.
A literature synthesis that unifies hallucination taxonomies across medical imaging modalities, finds general-purpose foundation models hallucinate less than specialized ones, and maps mitigation to FDA lifecycle frameworks.
Proposes bidirectional token-wise KL regularizer and visual-contrastive grounding objective to create fine-grained on-policy preference pairs for medical LVLMs by minimally editing model outputs.
MedSynapse-V proposes a latent memory evolution framework with meta-query prior retrieval, causal counterfactual refinement via RL, and intrinsic memory transition to improve diagnostic accuracy over chain-of-thought baselines in medical VLMs.
citing papers explorer
-
SHOVIR: A Benchmark for Evaluating Vision Shortcut Learning in Radiology Report Generation
SHOVIR benchmark uses region-level occlusion of chest X-rays to show that top-scoring radiology report generators often rely on shortcuts rather than on the visible evidence.
-
MM-Snowball: Evaluating and Mitigating Hallucination Snowballing in Multimodal Multi-Turn Dialogue
MM-Snowball benchmark diagnoses hallucination snowballing in multi-turn MLLM dialogues; CAVR mitigates it via dual visual rectification at representation and logit levels.
-
HalluCXR: Benchmarking and Mitigating Hallucinations in Medical Vision-Language Models for Chest Radiograph Interpretation
HalluCXR benchmark shows 61.9-82.3% hallucination rates across VLMs on MIMIC-CXR images, identifies patterns such as length-based risk and over-fabrication of common findings, and demonstrates ensemble mitigation that cuts fabrication by up to 84.8%.
-
Evaluating the Search Agent in a Parallel World
Mind-ParaWorld creates parallel worlds with atomic facts to evaluate search agents on future scenarios, showing they synthesize evidence well but struggle with collection, coverage, sufficiency judgment, and stopping decisions.
-
Detecting Clinical Hallucinations in LVLMs via Counterfactual Visual Grounding Uncertainty
A counterfactual visual grounding uncertainty method detects hallucinations in LVLMs on medical images, improving over baselines with interpretable evidence and cross-model transfer.
-
MedBench v5: A Dynamic, Process-Oriented, and Hallucination-Aware Benchmark for Clinical Multimodal Models
Strong medical-model task scores do not guarantee process stability: omission, contradiction, and delay mainly break contradiction detection, diagnosis updating, and hallucination self-correction while final evidence grounding can look fine.
-
Reducing Object Hallucination in LVLMs via Emphasizing Image-negative Tokens
Reweighting training emphasis toward image-negative tokens and filtering hallucinated data reduces object hallucination in LVLMs across three model variants.
-
MedVIGIL: Evaluating Trustworthy Medical VLMs Under Broken Visual Evidence
MedVIGIL provides a 300-case evaluation suite with 2556 probes that measures silent failures in medical VLMs under broken evidence, showing the best model at 69.2 on the composite score versus a human radiologist at 83.3.
-
Hallucination Detection and Correction in Medical VLMs via Counter-Evidence Verification
CoEV is a plug-and-play bidirectional verification method that maps text statements to visual evidence regions, assigns them to a four-quadrant factuality-grounding map, and uses this to detect and correct hallucinations in medical VLMs without retraining.
-
VIHD: Visual Intervention-based Hallucination Detection for Medical Visual Question Answering
VIHD detects hallucinations in medical MLLMs via visual dependency probing, token masking intervention, and calibrated semantic entropy on VQA benchmarks.
-
Mitigating Hallucinations in Large Vision-Language Models without Performance Degradation
MPD reduces hallucinations in LVLMs by 23.4% while retaining 97.4% of general capability through semantic disentanglement and selective parameter updates.
-
Mitigating Entangled Steering in Large Vision-Language Models for Hallucination Reduction
MESA reduces hallucinations in LVLMs via controlled selective latent intervention that preserves the original token distribution.
-
Hallucination in Medical Imaging AI: A Cross-Modality Analytical Framework for Taxonomy, Detection, and Mitigation under Regulatory Constraints
A literature synthesis that unifies hallucination taxonomies across medical imaging modalities, finds general-purpose foundation models hallucinate less than specialized ones, and maps mitigation to FDA lifecycle frameworks.
-
Analyzing and Improving Fine-grained Preference Optimization in Medical LVLMs
Proposes bidirectional token-wise KL regularizer and visual-contrastive grounding objective to create fine-grained on-policy preference pairs for medical LVLMs by minimally editing model outputs.
-
MedSynapse-V: Bridging Visual Perception and Clinical Intuition via Latent Memory Evolution
MedSynapse-V proposes a latent memory evolution framework with meta-query prior retrieval, causal counterfactual refinement via RL, and intrinsic memory transition to improve diagnostic accuracy over chain-of-thought baselines in medical VLMs.