Comparative evaluation of large language models for generating CAD-RADS 2.0-compliant diagnostic conclusions in cardiac CT reports

Key Points

• This research aims to assess the effectiveness of large language models in generating CAD-RADS 2.0-compliant diagnostic conclusions for cardiac CT reports.
• Comparison of various large language models in generating reports
• Evaluation of compliance with CAD-RADS 2.0 standards
• Assessment of reporting efficiency
• LLMs showed strong potential for automating structured reporting
• Significant improvements in reporting efficiency were observed
• Further optimization of LLMs is necessary before they can be integrated clinically