cs.CLOct 31, 2026

CheXPoT: Post-Hoc Visual Verification and Correction of VLM-Generated Radiology Reports

Authors: Joey Guo, Rayan Ansari, Alaa Youssef

Organizations: The Nueva School · Stanford Cardiovascular Institute · Department of Radiology, Stanford School of Medicine

Read PDF
Ask AI
Cite

Copy citation

DOINot available for this paper

Abstract

Vision-language models (VLMs) have shown promise for chest radiograph report generation, but adoption remains limited because of hallucinated findings that are not grounded in the image. Correcting these hallucinations and improving overall report quality remains a significant challenge. Prior work has addressed this problem by refining models with reinforcement learning, preference optimization, or chain-of-thought supervision, often at substantial compute cost. To address this gap, we introduce CheXPoT, a program-of-thought (PoT) framework that verifies each generated finding against image features and removes those it cannot ground. CheXPoT decomposes generated reports into predicates with RadGraph and an LLM, grounds these predicates against a DenseNet-121 and Grad-CAM visual evidence table, executes PoT verification code, and rewrites reports after removing unsupported predicates. On a 4,145-image held-out split, CheXPoT yields statistically significant precision gains on 8 of 14 CheXpert classes and F1 gains on 5, with no significant decreases in either metric. CheXPoT lifts micro-average precision from 19.2% (MedGemma-4B) to 46.8% (MedGemma-4B + CheXPoT) and micro-average F1 from 23.0% to 49.4%, absolute gains of 27.6% and 26.4% respectively. We also tested CheXPoT on other VLM report generators. On LLaVA-Med, CheXPoT lifts micro-average precision from 4.5% to 17.7% and micro-average F1 from 6.3% to 19.4%. On MAIRA-2, CheXPoT lifts micro-average precision from 15.6% to 42.1% and micro-average F1 from 21.0% to 45.3%. Our results suggest that inexpensive post-hoc visual verification may improve VLM trustworthiness by correcting unsupported findings from generated reports. CheXPoT lays a foundation for tools that help validate VLM-generated radiology reports and reduce unsupported findings. These innovations may subsequently help increase trust and adoption of VLM tools in Medicine.

More in this topic

VLM Hallucination Mitigation

Vision-language model hallucination mitigation keeps generated claims grounded in visual evidence.

See all 235 papers

Explore similar work

CardsList
Jul 21, 2026

DobicVLM: Aligning Chest X-Ray Report Generation with Clinically-Grounded Programmatic Rewards via Group Relative Policy Optimization

Medical imaging is a cornerstone of diagnostics, yet automated chest X-ray report generation struggles with structural adherence, anatomical completeness, and semantic faithfulness. We introduce DobicVLM, a vision-language model combining supervised fine-tuning on MedGemma-4B with Group Relative Policy Optimization (GRPO) and clinically-grounded programmatic rewards. Our approach uses interpretable, rule-based reward components; structural verification, anatomical checklist, semantic similarity, and length constraints to enforce clinical standards without neural reward models. Trained on 1,000 de-identified image-report pairs from a private clinical dataset (with ethics approval and compliance to local regulations), DobicVLM is evaluated via blinded expert review on 69 held-out cases. DobicVLM outperforms Gemini 2.5 Flash across the majority of criteria, achieving the highest impression accuracy (27.2%) and medical terminology (86.5%) compared to both Gemini 2.5 Flash and MedGemma 4B baselines, with minor trade-offs in completeness and referrals. This demonstrates GRPO's value for transparent alignment in resource-limited settings. Keywords: Vision-Language Models, Radiology Report Generation, Reinforcement Learning, Medical AI, GRPO

Thanni Adewuyi, Angelica Obayi, Andem Aniekan +10

Dobic Health · HelpMum Africa · University of Ibadan

Radiology Report GenerationRadiology VLMsRL for VLMs
AskPDF
Aug 4, 2026

CARE-X: Towards Clinically Useful Radiology VLMs with Auxiliary Supervision, Reward-Aligned Learning, and Tool-Augmented Measurement

A clinically useful chest X-ray system must go beyond fluent report generation: it should classify findings with tunable decision thresholds, localize them spatially, and derive the anatomical measurements upon which many diagnoses depend. Today's Vision-Language Models (VLMs) treat these as separate problems, if they address them at all, leaving a gap between what radiologists need and what generative models provide. We introduce CARE-X, a chest X-ray VLM that narrows this gap by unifying auxiliary discriminative supervision with reward-aligned generation. CARE-X augments its generative backbone with focal-loss classification and composite-loss grounding heads, co-trained alongside the language-modeling objective. This auxiliary supervision produces discriminative diagnostic predictions with tunable decision thresholds and precise spatial localization while also improving report quality, providing evidence that structured prediction and generation reinforce one another. Building on this foundation, Decoupled Clip and Dynamic Sampling Policy Optimization (DAPO) leverages task-specific reward signals for report generation, visual question answering (VQA), and spatial grounding, directly optimizing the clinical quality metrics that matter in practice. The result is state-of-the-art performance on the majority of metrics across four report-generation benchmarks, 94.0% VQA accuracy on ReXVQA (+6.0 pp over the next-best baseline), and generative spatial decoding that reaches near parity with dedicated detection heads. Separately, to address measurement-dependent diagnoses, we couple Qwen3-VL-4B-Instruct with native tool-calling capabilities for invoking deterministic measurement tools, while retaining full visual access to the image. This hybrid inference yields +43.6 pp average F1 over perception-only baselines across five measurement-dependent conditions.

Mercy Prasanna Ranjit, Anirban Porya, Sathvik Joel +8

Microsoft Research India · Medha AI, Narayana Health, India. · RTIICS, Narayana Health, India

Radiology Report GenerationChest X-Ray ClassificationRadiology VLMsVisual GroundingRL for VLMs
AskPDF
May 19, 2026

HalluCXR: Benchmarking and Mitigating Hallucinations in Medical Vision-Language Models for Chest Radiograph Interpretation

Vision-language models (VLMs) are increasingly used for medical image interpretation, yet they frequently hallucinate, generating clinically plausible but factually incorrect findings that pose direct patient safety risks. We introduce HalluCXR, a benchmark evaluating six architecturally diverse VLMs across 856 stratified MIMIC-CXR chest radiographs and three query types, yielding 15,408 model evaluations. An eight-category hallucination taxonomy with clinical severity ratings and a two-layer detection pipeline are validated against 250 human annotations (auto-detection F1=0.959; LLM judge F1=0.907). We find that 61.9--82.3% of outputs contain hallucinations, with clinically dangerous errors in up to 80.2%. Three key patterns emerge: normal radiographs paradoxically attract the most severe hallucinations, common findings are systematically over-fabricated while rare findings go under-detected, and response length alone predicts hallucination risk (AUC up to 0.908). A six-model ensemble reduces fabrication by up to 84.8% at the cost of increased omission; a three-model subset retains comparable performance at half the cost. These results establish that hallucination auditing, verbosity-based risk monitoring, and ensemble-based safety layers are prerequisites for clinical deployment.

Haoyu Wang, Zitong Li

Department of Biostatistics & Health Informatics, Institute of Psychiatry, Psychology & Neuroscience, King’s College London, London, UK

Vision-Language ModelsAudio-Language Model Hallucination DetectionHealthcareHallucination in Language ModelsVLM HallucinationMedical VLMs+2
AskPDF