A large language model (LLM) significantly outperforms RadLex in expanding terms for radiology report language standardization.
Key Details
- 1Study published in American Journal of Roentgenology compared LLM to RadLex for term expansion in radiology reports.
- 2LLM (Gemini 2.0 Flash Thinking) generated 208,465 additional variants and 69,918 synonyms beyond RadLex's expansion.
- 3LLM expansion improved lexical coverage rate to 81.9% vs. RadLex's 67.5%.
- 4Semantic recall improved to 81.6% (LLM) versus 64% (RadLex), with slightly lower precision (94.8% vs 100%).
- 5F1 score was higher for LLM expansion (0.91) compared to RadLex (0.86).
- 6Study used chest CT reports from five international datasets.
Why It Matters
Automating terminology expansion with LLMs can enhance the accuracy and scalability of natural language processing in radiology, aiding standardized reporting, AI model development, and multi-center research.
Source
AuntMinnie
Related News
•Radiology Business
AI Model Aims to Detect Placenta Accreta Earlier in Pregnancy
Researchers present a new AI model to help diagnose placenta accreta spectrum before delivery.
•Radiology Business
AI Analyzes Mammograms to Assess Post-Surgery Breast Cancer Recurrence Risk
A new AI tool predicts likelihood of breast cancer recurrence after surgery using pre-operative mammograms.
•Radiology Business
FDA Clears Median's Eyonis AI for CT Lung Cancer Screening
Median Technologies' Eyonis LCS, an AI tool for low-dose CT lung cancer screening, receives FDA clearance.