Assessing GPT-4 performance and alignment with human expertise for text-based MRI protocol assignment

Key Points

• GPT-4 demonstrated high accuracy in MRI protocol assignment, closely matching human expertise.
• The performance metrics indicated over 85% alignment with human-generated MRI protocol decisions.
• Analysis using text-based assessments highlighted GPT-4's ability to understand complex medical protocols effectively.
• This study underscores the potential benefits of AI in assisting medical professionals, particularly in MRI tasks.