Researchers have developed an LLM-based system to filter pathological findings from clinical CT scan reports. This method allows for the creation of healthier reference cohorts from over 350,000 CT examinations. The system uses five LLMs to identify and resolve disagreements on abnormality candidates, enabling the establishment of comprehensive whole-body reference charts for 106 anatomical structures. These charts detail organ volume and tissue attenuation, accounting for factors like age, sex, and contrast enhancement, and can aid in standardized quantitative phenotyping and screening research. AI
Summary written by gemini-2.5-flash-lite from 1 source. How we write summaries →
IMPACT Novel LLM application in medical imaging analysis could improve diagnostic accuracy and research capabilities.
RANK_REASON This is a research paper published on arXiv detailing a novel method for processing clinical data. [lever_c_demoted from research: ic=1 ai=1.0]