PromptRad: Knowledge-Enhanced Multi-Label Prompt-Tuning for Low-Resource Radiology Report Labeling

2026-05-19

AI Summary

This work addresses the challenge of accurately identifying multi-label clinical findings from unstructured radiology reports in data-scarce clinical settings. The authors propose a novel approach that reformulates multi-label classification as a masked language modeling task, leveraging UMLS synonyms to construct a knowledge-enhanced, multi-token verbalizer and employing prompt-based fine-tuning of a pre-trained language model—eliminating the need for an additional classification layer. This method uniquely integrates knowledge-augmented verbalization with prompt tuning, substantially reducing reliance on labeled data and effectively handling complex negation expressions. Remarkably, with only 32 annotated samples, the approach outperforms dictionary-based and standard fine-tuning baselines on liver CT reports, achieving performance close to that of GPT-4 while more precisely capturing negation patterns.

Abstract

Automatic report labeling facilitates the identification of clinical findings from unstructured text and enables large-scale annotation for medical imaging research. Existing rule-based labelers struggle with the diverse descriptions in clinical reports, while fine-tuning pre-trained language models (PLMs) requires large amounts of labeled data that are often unavailable in clinical settings. In this paper, we propose PromptRad, a knowledge-enhanced multi-label \textbf{prompt}-tuning approach for \textbf{rad}iology report labeling under low-resource settings. PromptRad reformulates multi-label classification as masked language modeling and incorporates synonyms from the UMLS Metathesaurus into a multi-word verbalizer to enrich category representations. By fine-tuning the PLM without additional classification layers, PromptRad requires substantially less labeled data than conventional fine-tuning. Experiments on liver CT reports show that PromptRad outperforms dictionary-based and fine-tuning baselines with only 32 labeled training examples, and achieves competitive performance with GPT-4 despite using a much smaller model. Further analysis demonstrates that PromptRad captures complex negation patterns more effectively than existing methods, making it a promising solution for report labeling in data-scarce clinical scenarios. Our code is available at https://github.com/ila-lab/PromptRad.

Problem

• radiology report labeling
• low-resource
• multi-label classification
• clinical text
• automatic annotation

Innovation

• prompt-tuning
• multi-label classification
• knowledge-enhanced
• low-resource learning
• radiology report labeling