Clinical Implementation of AI for Pulmonary Embolism Detection in over 30,000 CT Pulmonary Angiography Examinations

Key Points

• This research aims to evaluate the effectiveness of AI in detecting pulmonary embolism through CT pulmonary angiography.
• Large-scale deployment of AI across over 30,000 CT pulmonary angiography examinations.
• Comparison of AI detection results with radiologist interpretations.
• Assessment of scenarios featuring discordances between AI and expert reviews.
• AI demonstrated high concordance with radiologists, particularly in cases of discordance.
• Expert oversight revealed complementary roles between radiologists and AI.
• Specific scenarios highlighted divergences in detection capabilities between AI and radiologists.