LLM-assisted label cleaning improves chest CT dataset accuracy

Researchers have developed a method using large language models (LLMs) to improve the accuracy of labels in large-scale medical imaging datasets. By comparing existing labels in the CT-RATE chest CT dataset with labels generated by GPT-5.4, they identified instances of label-report discordance. Radiologist adjudication supported the LLM-derived labels in a significant majority of cases, suggesting that LLM-assisted cleaning can enhance the quality of public imaging datasets for future research. AI

IMPACT Enhances the quality and reliability of medical imaging datasets, potentially accelerating AI research and development in healthcare.

RANK_REASON Academic paper detailing a new methodology for data cleaning using LLMs. [lever_c_demoted from research: ic=1 ai=1.0]

Read on arXiv cs.AI →

• arXiv
• CT-RATE
• GPT-5.4
• Hugging Face
• Large Language Model
• Microsoft Azure AI Foundry

• paper
• model release

AI-generated summary · Google Gemini · from 1 sources. How we write summaries →