Radiologist-AI Combo Has Highest Potential to Improve Pulmonary Embolism Detection, Experts Charge

Why It Matters

Combining AI with radiologist expertise dramatically boosts detection accuracy, addressing the growing radiology workload and shortage while potentially saving lives from missed pulmonary embolisms.

Key Takeaways

• AI identified 85% of PE cases before radiologist confirmation
• Radiologist‑AI combo achieved 99.2% sensitivity for PE detection
• Radiologists corrected 88.7% of AI‑radiologist disagreements
• Concordance higher for negative exams (98.2%) than positives (93.8%)
• Study covered 32,500 CTPA exams from 29,000 adults

Pulse Analysis

Pulmonary embolism remains a leading cause of sudden death, with over 300,000 U.S. fatalities annually. Traditional diagnosis relies on CT pulmonary angiography interpreted by radiologists, but escalating imaging volumes and a global radiologist shortage strain timely reads. Artificial intelligence offers a promising adjunct, capable of rapidly scanning images for clot signatures, yet concerns linger about false positives and the need for clinical oversight.

The Neiman Health Policy Institute’s retrospective analysis of 32,500 CTPA scans applied Aidoc’s AI algorithm across a large, real‑world hospital network. AI flagged 85% of positive PE cases, and when radiologists reviewed AI‑highlighted studies, overall sensitivity rose to 99.2%. Discrepancies were resolved in favor of radiologists 88.7% of the time, underscoring that AI serves best as a triage tool rather than a replacement. Notably, agreement rates were highest for negative exams (98.2%) and for acute, central emboli, the most clinically urgent presentations.

These results signal a shift toward hybrid diagnostic pathways where AI accelerates case prioritization and radiologists focus on nuanced interpretation. Hospitals can expect reduced turnaround times, lower missed‑diagnosis risk, and more efficient allocation of scarce radiology talent. As vendors refine algorithms and regulatory frameworks evolve, broader adoption of AI‑informed radiology could become a standard of care, reshaping workflow economics and patient outcomes across the healthcare system.