Abstract
Pulmonary embolism (PE) is a potentially fatal condition requiring prompt and accurate diagnosis. Computed tomographic pulmonary angiography (CTPA) is the gold standard for PE detection, but its interpretation is time-intensive and subject to human error. Recent advancements in artificial intelligence (AI), particularly in machine learning (ML) and deep learning (DL) algorithms, offer promising tools to enhance diagnostic efficiency and accuracy. This systematic review and meta-analysis evaluated the diagnostic performance of FDA-approved ML algorithms for detecting acute PE on CTPA. A comprehensive literature search across six databases identified six retrospective studies encompassing 9,102 CTPA scans, all of which assessed either the Aidoc or CINA-PE algorithms. Pooled sensitivity and specificity were 93% (95% CI: 88%-95%) and 98% (95% CI: 93%-100%), respectively, indicating high diagnostic accuracy across real-world datasets. All included studies demonstrated low risk of bias according to Quality Assessment of Diagnostic Accuracy Studies-2 (QUADAS-2) assessments. These findings support the integration of FDA-approved ML tools into radiological workflows as adjuncts to reduce diagnostic errors and improve triage. However, prospective trials are needed to assess their impact on clinical decision-making, workflow efficiency, cost-effectiveness, and patient outcomes.