Improving Automatic Summarization of Radiology Reports through Mid-Training of Large Language Models

2026-02-28

AI Summary

This work proposes a three-stage “pretraining–intermediate training–fine-tuning” paradigm to alleviate physician workload and enhance factual consistency and accuracy in automatic summarization of radiology reports. Following general-domain pretraining, the approach introduces an intermediate training phase tailored to the radiology subdomain, effectively mitigating cold-start issues and improving few-shot learning capabilities. Built upon the GatorTronT5 architecture, the model undergoes intermediate training on large-scale clinical text from UF Health and is subsequently fine-tuned on the OpenI and MIMIC-CXR datasets. The resulting model, GatorTronT5-Radio, significantly outperforms existing baselines on both ROUGE-L and RadGraph-F1 metrics, demonstrating its effectiveness in generating high-quality radiology summaries that accurately reflect medical facts.

Technology Category

• Natural Language Processing: Summarization
• Machine Learning: Large Multimodal Models (LMMs)
• Computer Vision: Multi-modal Vision

Application Category

• Search and Retrieval-Augmented AI: Retrieval-Augmented Generation (RAG) and multi-modal RAG
• User Modeling, Personalization and Recommendation: Fairness-aware retrieval and ranking
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs

Abstract

Automatic summarization of radiology reports is an essential application to reduce the burden on physicians. Previous studies have widely used the"pre-training, fine-tuning"strategy to adapt large language models (LLMs) for summarization. This study proposed a subdomain adaptation through a mid-training method to improve summarization. We explored three adaptation strategies: (1) general-domain pre-training, (2) clinical-domain pre-training, and (3) clinical-domain pre-training followed by subdomain mid-training. We developed models using large-scale clinical text from the University of Florida (UF) Health and conducted mid-training and fine-tuning experiments using widely used benchmark datasets including OpenI and MIMIC-CXR. The experimental results show that the mid-trained model, GatorTronT5-Radio, achieved the best performance, outperforming models without mid-training in both text-based measures (ROUGE-L) and factuality measures (RadGraph-F1). Our mid-training methods also demonstrate better few-shot learning and could alleviate the"cold start"problem reported in previous studies as a learning barrier. Our findings support the use of"pre-training, mid-training, fine-tuning,"instead of the widely used direct fine-tuning strategy.

Problem

• Automatic Summarization
• Radiology Reports
• Large Language Models
• Subdomain Adaptation
• Mid-Training

Innovation

• mid-training
• radiology report summarization
• large language models
• subdomain adaptation
• factuality evaluation