An LLM-based chart summarization tool safely reduced radiologist study review time by 17%.
Key Details
- 1Stanford researchers piloted an LLM-based chart summarizer integrated with Epic Imaging Insights.
- 2102 LLM-generated summaries were evaluated by radiologists using a structured survey.
- 3All were considered safe, with 86% rated clear and 85% accurate, but only 70% deemed complete and 67% relevant.
- 4Over 133,000 CT and MRI exams were tracked before and after tool deployment.
- 5Median review time dropped from 11.3 to 9.4 minutes per case after implementation, a 17% reduction.
- 6Most issues involved missing key clinical details in summaries.
Why It Matters

Source
AuntMinnie
Related News

Two-Stage LLM Pipeline Boosts Patient-Friendly Radiology Reports
A two-stage LLM process significantly improves the readability and accuracy of radiology reports for patients.

Deep Learning on PSMA-PET Scans Predicts Survival in Prostate Cancer
Deep learning models applied to PSMA-PET scans outperform existing tools in predicting two-year survival for metastatic prostate cancer.

RSNA 2026 Highlights: Photon-Counting CT and AI Innovations in Clinical Imaging
RSNA 2026 will spotlight advances in CT imaging, featuring photon-counting CT, deep learning image reconstruction, and emerging applications that enhance resolution, lower dose, and add quantitative value to diagnostics.