SURE-Med: Systematic Uncertainty Reduction for Enhanced Reliability in Medical Report Generation

2025-08-03

AI Summary

Medical report generation (MRG) suffers from three interrelated uncertainties—visual (view-label noise), distributional (long-tailed disease bias), and contextual (historical report hallucination)—severely undermining clinical reliability. To address this, we propose the first unified framework systematically mitigating all three uncertainties: (1) Frontal-Aware View Repair Resampling corrects erroneous view annotations via anatomically informed resampling; (2) Token-Sensitive Learning enhances modeling fidelity for rare-disease–associated tokens through adaptive token-level weighting; and (3) Contextual Evidence Filtering suppresses factual hallucinations by cross-validating generated content against historical reports and filtering spurious assertions. Integrating feature reweighting with multi-source evidence verification, our method significantly improves report accuracy and clinical trustworthiness on MIMIC-CXR and IU-Xray. This work establishes a new paradigm for trustworthy, uncertainty-aware AI decision support in clinical radiology.

Technology Category

• Machine Learning: Calibration & Uncertainty Quantification
• Reasoning under Uncertainty: Uncertainty Representations
• Computer Vision: Multi-modal Vision

Application Category

• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• User Modeling, Personalization and Recommendation: Fairness-aware retrieval and ranking
• Economics, Online Markets and Human Computation: Trust and reliance of crowd workers and data experts on GenAI

Abstract

Automated medical report generation (MRG) holds great promise for reducing the heavy workload of radiologists. However, its clinical deployment is hindered by three major sources of uncertainty. First, visual uncertainty, caused by noisy or incorrect view annotations, compromises feature extraction. Second, label distribution uncertainty, stemming from long-tailed disease prevalence, biases models against rare but clinically critical conditions. Third, contextual uncertainty, introduced by unverified historical reports, often leads to factual hallucinations. These challenges collectively limit the reliability and clinical trustworthiness of MRG systems. To address these issues, we propose SURE-Med, a unified framework that systematically reduces uncertainty across three critical dimensions: visual, distributional, and contextual. To mitigate visual uncertainty, a Frontal-Aware View Repair Resampling module corrects view annotation errors and adaptively selects informative features from supplementary views. To tackle label distribution uncertainty, we introduce a Token Sensitive Learning objective that enhances the modeling of critical diagnostic sentences while reweighting underrepresented diagnostic terms, thereby improving sensitivity to infrequent conditions. To reduce contextual uncertainty, our Contextual Evidence Filter validates and selectively incorporates prior information that aligns with the current image, effectively suppressing hallucinations. Extensive experiments on the MIMIC-CXR and IU-Xray benchmarks demonstrate that SURE-Med achieves state-of-the-art performance. By holistically reducing uncertainty across multiple input modalities, SURE-Med sets a new benchmark for reliability in medical report generation and offers a robust step toward trustworthy clinical decision support.

Problem

• Reduces visual uncertainty from noisy view annotations
• Addresses label distribution uncertainty in long-tailed diseases
• Mitigates contextual uncertainty to prevent factual hallucinations

Innovation

• Frontal-Aware View Repair Resampling corrects annotation errors
• Token Sensitive Learning reweights underrepresented diagnostic terms
• Contextual Evidence Filter validates prior information to suppress hallucinations