GPT-4o, with prompt engineering, selected optimal abdominal/pelvic CT protocols more accurately than radiologists without increasing inappropriate selections.
Key Details
- 1Study evaluated 1,448 abdominal and pelvic CT exams between Jan-June 2024.
- 2GPT-4o with detailed prompting selected optimal protocols 96.2% of the time, compared to 88.3% for radiologists (p<0.001).
- 3Rates of inappropriate protocols were similarly low: 1.3% (GPT-4o) vs. 2.4% (radiologists), not statistically significant (p=0.21).
- 4Fine-tuning GPT-4o offered no performance increase over meticulous prompting (both 96.2%).
- 5Performance in protocol matching was consistent across training levels (radiologist, fellow, resident).
Why It Matters
Automating protocol assignment with LLMs like GPT-4o could streamline imaging workflow, freeing radiologists for more complex tasks and reducing the risk of workflow interruptions and diagnostic errors. The approach may allow for adaptable, institution-specific automation without intensive retraining.
Source
AuntMinnie
Related News
•HealthExec
New Report Highlights Clinical AI Performance, Sustainability, and Adoption Challenges
A multi-institutional review details key challenges, progress, and sustainability concerns in deploying clinical AI in real-world healthcare settings.
•Cardiovascular Business
FDA Clears AI Platform for Comprehensive Cardiac Risk Assessment on CT
HeartLung Corporation's AI-CVD receives FDA clearance for opportunistic multi-condition screening on routine chest CT scans.
•AuntMinnie
LLM Boosts Terminology Expansion in Radiology Reports Over RadLex
A large language model (LLM) significantly outperforms RadLex in expanding terms for radiology report language standardization.