Radiology Report Generation for Low-Quality X-Ray Images

arXiv cs.CV / 4/14/2026

💬 OpinionIdeas & Deep AnalysisModels & Research

共有:

Key Points

The paper addresses how existing vision-language models for radiology report generation fail when X-ray inputs are noisy or low-quality, causing substantial real-world performance degradation.
It introduces an Automated Quality Assessment Agent (AQAA) to detect low-quality samples in the MIMIC-CXR dataset and establishes a dedicated Low-quality Radiology Report Generation (LRRG) benchmark.
To reduce degradation-induced performance shifts, the authors propose a Dual-loop Training Strategy using bi-level optimization and gradient consistency to learn quality-agnostic diagnostic features.
Experiments show the approach mitigates the drop in report generation quality as image quality deteriorates, and the authors plan to release code and data after acceptance.

Abstract

Vision-Language Models (VLMs) have significantly advanced automated Radiology Report Generation (RRG). However, existing methods implicitly assume high-quality inputs, overlooking the noise and artifacts prevalent in real-world clinical environments. Consequently, current models exhibit severe performance degradation when processing suboptimal images. To bridge this gap, we propose a robust report generation framework explicitly designed for image quality variations. We first introduce an Automated Quality Assessment Agent (AQAA) to identify low-quality samples within the MIMIC-CXR dataset and establish the Low-quality Radiology Report Generation (LRRG) benchmark. To tackle degradation-induced shifts, we propose a novel Dual-loop Training Strategy leveraging bi-level optimization and gradient consistency. This approach ensures the model learns quality-agnostic diagnostic features by aligning gradient directions across varying quality regimes. Extensive experiments demonstrate that our approach effectively mitigates model performance degradation caused by image quality deterioration. The code and data will be released upon acceptance.