Combining AI with radiologists substantially improves detection of pulmonary embolism (PE) on CT pulmonary angiography (CTPA) scans, new research shows.
Key Details
- 1Researchers retrospectively analyzed 32,500 CTPA exams from over 29,000 adults (2021-2023).
- 2AI and radiologists agreed on PE diagnosis in 97.8% of cases.
- 385% of positive PE cases were flagged by AI and confirmed by radiologists; 15% were found only by radiologists.
- 4Radiologists were correct in 88.7% of the disagreements; AI in 11.3%.
- 5AI-informed radiologists achieved 99.2% sensitivity for PE detection.
- 6Concordance was highest for acute and central emboli, which have the most clinical urgency.
Why It Matters
This study demonstrates that integrating AI with radiologist expertise can improve both the efficiency and accuracy of PE detection, especially for urgent cases. It highlights the necessity of human oversight, illustrating the best results stem from collaborative use rather than AI or radiologist alone.
Source
Radiology Business
Related News
•Radiology Business
AI-Generated Report Summaries Improve Patient Understanding at Emory Radiology
Emory University demonstrated that LLM-based tools can enhance patient comprehension of radiology reports.
•AuntMinnie
Cardiac MRI Plus AI Boosts Heart Risk Prediction Accuracy
A random-forest machine learning model leveraging cardiac MRI and patient health history improves prediction of major adverse cardiovascular events.
•AuntMinnie
SIIM Study: LLMs Excel in Numerical Radiology Tasks
At SIIM 2026, LLMs demonstrated high accuracy on radiology-based numerical tasks, particularly in extraction and judgment tests.