Radiology Report Generation for Low-Quality X-Ray Images

Hongze Zhu; Chen Hu; Jiaxuan Jiang; Hong Liu; Yawen Huang; Ming Hu; Tianyu Wang; Zhijian Wu

cs.CVApr 11, 2026

Radiology Report Generation for Low-Quality X-Ray Images

Authors: Hongze Zhu, Chen Hu, Jiaxuan Jiang, Hong Liu, Yawen Huang, Ming Hu, Tianyu Wang, Zhijian Wu, +1 more

Organizations: Westlake University · Tencent Jarvis Lab

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.

Figures & tables

Figure 1: Performance under quality degradation. We evaluate R2GenGPT Wang et al. (2023c) across quality regimes and observe consistent declines in report generation performance as image quality degrades from standard to mildly suboptimal and severely suboptimal .

Figure 2: Retake pairs illustrate acquisition-related quality shift. A low-quality pre-retake CXR can degrade report generation, while the post-retake CXR improves alignment with the reference report.

Figure 3: Overview of our quality assessment agent. We first extract no-reference IQA scores and an exposure deviation index from each CXR, then feed these priors with a CEC/EUR16260-aligned prompt into a VLM to classify image quality into three levels: standard, mildly suboptimal, and severely suboptimal.

Condition	Mildly ( $N=32{,}826$ )	Severely ( $N=22{,}840$ )
Enlarged Cardiomediastinum	2,019 (6.15%)	1,481 (6.48%)
Cardiomegaly	9,631 (29.34%)	6,471 (28.33%)
Lung Opacity	11,055 (33.68%)	8,870 (38.84%)
Lung Lesion	1,044 (3.18%)	647 (2.83%)
Edema	5,126 (15.62%)	3,974 (17.40%)
Consolidation	1,569 (4.78%)	1,343 (5.88%)

Table 1: Distribution of labels in Mildly (

N=32,826

) and Severely (

N=22,840

) Suboptimal datasets.

Quality Regime	Partition Statistics
Quality Regime	Train	Val	Test
Standard ( $\mathcal{D}_{std}$ )	104,827	857	1,780
Mildly Suboptimal ( $\mathcal{D}_{mild}$ )	28,791	240	515
Severely Suboptimal ( $\mathcal{D}_{severe}$ )	19,189	186	514
Auxiliary Low-Quality	–	–	669
Total	152,807	1,283	3,478

Table 2: Statistics of the constructed quality regimes.

\mathcal{D}_{std}

represents ideal stationary scans.

\mathcal{D}_{mild}

and

\mathcal{D}_{severe}

denote scans with graded quality degradation.

Figure 4: Overview of our quality assessment agent. We compute no-reference IQA signals and an exposure deviation index for each CXR, then use a VLM with a CEC/EUR16260-aligned prompt to grade images into three levels: standard, mildly suboptimal, and severely suboptimal.

		NLG METRICS				CE METRICS
Benchmarks	Methods	BLEU-1	BLEU-4	METEOR	ROUGE-L	Precision	Recall	F1 Score	RaTEScore	RadGraph
$T_{std}$	GPT5.2 Achiam et al. (2023)	0.137	0.006	0.108	0.109	0.215	0.203	0.185	0.481	0.180
	EKAGen Bu et al. (2024)	0.422	0.120	0.158	0.271	0.385	0.278	0.294	0.534	0.284
	Mamba_Xray Wang et al. (2025a)	0.398	0.121	0.172	0.270	0.331	0.258	0.277	0.540	0.280
	Qwen3VL4B Yang et al. (2025)	0.087	0.009	0.111	0.115	0.215	0.228	0.183	0.518	0.155
	Qwen3VL4B-sft Yang et al. (2025)	0.243	0.063	0.141	0.212	0.332	0.247	0.252	0.538	0.271

Table 3: Performance comparison on the four benchmarks. CE Metrics now encompass advanced clinical alignment scores. Where sft denotes supervised fine-tuning, and DTS refers to the proposed dual-loop training strategy. We adopt the color scheme: Best and Second Best .

Explore similar work

Multimodal Large Language Model driven Radiology Report Generation with Clinical Knowledge Enhancement
Mar 11, 2024Miaojing Shi, Tianyu Cen, Zijie Yue +3Radiology Report Generation Radiology VLMs
College of Electronic and Information Engineering, Tongji University, Shanghai, China
CliQ-RRG: Clinical-Knowledge Guided Disease-aware Visual-Textual Alignment for QA-Style Radiology Report Generation
Oct 9, 2026Chandravardhan Singh Raghaw, Shahid Shafi Dar, Arnav Nishith Sharma +2Radiology Report Generation Medical VLMs
Indian Institute of Technology Indore, Indore, India · University of Texas at Dallas, USA · University of Wisconsin - Madison, USA
DRRG: A Discrete Diffusion Framework for Radiology Report Generation
Aug 25, 2026Shaoyang Zhoua, Yingshu Li, Yunyi Liu +4Radiology Report Generation Radiology VLMs
School of Electrical and Computer Engineering, The University of Sydney, Sydney, NSW, 2006, Australia · Department of Radiology, The First Affiliated Hospital of Soochow University, Suzhou, Jiangsu, 215006, China · School of Computer Science, The University of Adelaide, Adelaide, SA, 5005, Australia +1

Mar 11, 2024J/K move · Enter open · S save

Multimodal Large Language Model driven Radiology Report Generation with Clinical Knowledge Enhancement

Miaojing Shi, Tianyu Cen, Zijie Yue, Meng Wei+2

College of Electronic and Information Engineering, Tongji University, Shanghai, China

Radiology Report Generation Radiology VLMs Multimodal Large Language Models

Open paper AskPDF arXiv

Radiology report generation (RRG) has attracted significant attention due to its potential to reduce the workload of radiologists. The performance of current RRG approaches remains unsatisfactory against clinical standards. This paper introduces a novel RRG method, MLLM-RRG, that integrates multimodal large language models (MLLMs) with various types of clinical knowledge to generate accurate and comprehensive chest X-ray reports. Our method first designs a referring anatomical feature extractor that leverages anatomical knowledge to analyze different regions of the chest X-ray image and extract visual features without explicitly detecting regions. Next, based on the MLLM's decoder, we develop a multimodal report generator that leverages multimodal prompts constructed from dedicated visual features and textual instructions to produce the radiology report in an auto-regressive way. Finally, we introduce a disease-oriented clinical classification and alignment scheme in a multi-task learning manner to leverage disease knowledge to better preserve the clinical relevance among the generated reports. Once the model is trained, we also introduce a novel clinical quality reinforcement learning strategy to enhance the MLLM with report knowledge, further refining the tones of the generated reports towards radiologists. Extensive experiments on the MIMIC-CXR and IU X-Ray datasets demonstrate the superiority of our method over the state of the art. Our codes will be available at https://github.com/viscom-tongji/MLLM-RRG.

Abstract

Explore similar work

Multimodal Large Language Model driven Radiology Report Generation with Clinical Knowledge Enhancement

CliQ-RRG: Clinical-Knowledge Guided Disease-aware Visual-Textual Alignment for QA-Style Radiology Report Generation

DRRG: A Discrete Diffusion Framework for Radiology Report Generation