Positive-Unlabeled Preference Optimization For Chest X-ray Report Generation

2026-08-05

AI Summary

This work addresses the under-reporting problem in radiology report generation caused by missing-label noise in training data—such as unannotated pathological findings—by proposing PU-DPO, a novel framework that integrates positive-unlabeled (PU) learning into preference optimization for medical report generation. PU-DPO treats unmentioned pathologies as unlabeled rather than negative samples and constructs edited contrastive response pairs to provide effective supervision signals through Direct Preference Optimization (DPO). Experimental results on multiple chest X-ray benchmarks demonstrate that PU-DPO significantly improves pathology detection rates and the recovery of hidden positive cases, exhibiting greater robustness to missing-label noise compared to existing methods.

Abstract

Vision-Language Models (VLMs) for radiology report generation are typically trained on retrospective clinical reports, which suffer from omission noise: clinically present findings are left unreported due to the omission of subtle findings. For example, prior studies show that cardiomegaly may be omitted from ICU chest X-ray reports when the imaging request is focused on monitoring support device placement. As a result, models trained with standard approaches inherit these omissions, learning to under-report findings themselves. We propose PU-DPO, a preference optimization framework to prevent omission noise from corrupting the preference signal. We reformulate the objective under a positive-unlabeled (PU) learning framework, treating absent mentions as unlabeled rather than truly negative. Our framework provides preference supervision using constructed contrastive pairs, generated using edits to model responses, producing variants that explicitly mention or omit a specific finding. Generated responses that mention the finding are naturally preferred in the context of visual evidence. Across semi-synthetic experiments and analyses on real-world chest radiograph benchmarks where adjudicated labels are available, PU-DPO yields consistent gains in detection rates and recovery of hidden positives across multiple pathologies, and is more robust to omission noise than prior approaches.

Problem

• omission noise
• chest X-ray report generation
• positive-unlabeled learning
• radiology report
• under-reporting

Innovation

• Positive-Unlabeled Learning
• Preference Optimization
• Omission Noise
• Chest X-ray Report Generation
• Vision-Language Models