AI-Based CTPA Stratification Achieves AUC 0.848 for High-Risk Pulmonary Embolism

CTPA-based AI achieved AUC 0.848 for classifying high-risk pulmonary embolism. Automated ventricular and arterial measurements agreed well with radiologists: MPAD ICC 0.94, PAOI 0.80. RVD/LVD ratio >1.2 yielded AUC 0.771.

• Single-center retrospective analysis of 115 patients diagnosed with PE; AI measurements compared with manual reads by two radiologists.
• AI model using all covariates classified high-risk PE with AUC 0.848, without external validation.
• Automated RVD/LVD ratio >1.2 alone identified high-risk PE with AUC 0.771.

Automated summary

RadPigeon summaries are original and for information only. They are not clinical advice.