A Disease Labeler for Chinese Chest X-Ray Report Generation

TL;DR

The work tackles the lack of Chinese chest X-ray report data by introducing a Chinese chest X-ray report disease labeler that combines a dual BERT encoder with a hierarchical disease–body-part label relation. This labeler enables large-scale CCXRD construction (51,262 images, 47,886 reports) and demonstrates superior disease-label annotation on expert-annotated data, outperforming English-focused baselines and GPT-based approaches. The approach highlights the importance of domain-specific pretraining and structured label relations for reliable clinical text labeling, with practical impact on dataset creation and evaluation of Chinese CXR report generation models. The CCXRD resource and labeling methodology offer a path toward robust Chinese medical NLP tools and more accurate clinical report generation in Chinese contexts.

Abstract

In the field of medical image analysis, the scarcity of Chinese chest X-ray report datasets has hindered the development of technology for generating Chinese chest X-ray reports. On one hand, the construction of a Chinese chest X-ray report dataset is limited by the time-consuming and costly process of accurate expert disease annotation. On the other hand, a single natural language generation metric is commonly used to evaluate the similarity between generated and ground-truth reports, while the clinical accuracy and effectiveness of the generated reports rely on an accurate disease labeler (classifier). To address the issues, this study proposes a disease labeler tailored for the generation of Chinese chest X-ray reports. This labeler leverages a dual BERT architecture to handle diagnostic reports and clinical information separately and constructs a hierarchical label learning algorithm based on the affiliation between diseases and body parts to enhance text classification performance. Utilizing this disease labeler, a Chinese chest X-ray report dataset comprising 51,262 report samples was established. Finally, experiments and analyses were conducted on a subset of expert-annotated Chinese chest X-ray reports, validating the effectiveness of the proposed disease labeler.

Figures (3)

• Figure 1: Overall Architecture for the Chinese Chest X-ray Report Disease Labeler
• Figure 2: Hierarchical Labels Relationship Diagram
• Figure 3: ChatGPT Prompts and the Corresponding Application Workflow