Rapid PE Triage and Notification (PETN)

Rapid PE Triage and Notification (PETN) is an AI-powered software designed to enhance clinical workflow by rapidly analyzing adult CTPA images for suspected central pulmonary embolism, thereby enabling prioritized review by clinicians. The system utilizes a machine learning algorithm, validated on diverse datasets from multiple scanner types with demonstrated sensitivity and specificity, to provide notifications based on detected findings, aiming to expedite the identification of critical cases without altering original diagnostic images.

Rapid PE Triage and Notification (PETN) is a radiological computer-aided triage and notification software designed to assist hospital networks and trained clinicians in managing workflow for adult patients suspected of central pulmonary embolism (PE). It analyzes computed tomography pulmonary angiography (CTPA) images, specifically single-energy exams, to flag potential positive PE findings, thereby aiding in the prioritization of cases for review. The system employs an artificial intelligence algorithm that operates on a server or standalone desktop, running in parallel with the standard image interpretation process. Clinicians receive notifications for cases with suspected findings, which include compressed preview images intended for informational and prioritization purposes only, not for direct diagnosis, and the original medical images remain unaltered. The AI/ML algorithm underlying Rapid PETN was developed using a dataset of 600 CTPA cases, balanced between positive and negative PE findings, for training (480 cases) and development validation (120 cases), achieving a sensitivity of 0.98 and specificity of 0.97 in development. Further specificity testing was conducted on an additional 276 negative cases. Performance validation of the final device was conducted on 306 CTPA cases, with ground truth established by a consensus of three experts. This validation demonstrated a sensitivity of 0.96 (95% CI: 0.92, 0.98) and a specificity of 0.89 (95% CI: 0.83, 0.93) in identifying central PE. The system also achieved a mean processing time of 2.64 minutes (95% CI: 2.34-4.80 min), with negligible notification delay using wireless and cellular communications, and has been validated for use with CTPA images acquired from GE, Siemens, and Toshiba scanners.

AI Regulatory Concerns

Powered by

Jev reads the full FDA letter and returns calibrated concern probabilities per regulatory category. These are AI estimates, not regulatory advice.

• Q1. How does the Rapid PETN algorithm differentiate central pulmonary emboli from more peripheral emboli, and what is its performance for detecting segmental or subsegmental PE if any?; What specific features or image characteristics does the AI/ML model primarily rely on to identify suspected central PE, and how robust is it to variations in contrast timing or image noise?; Given the performance validation showed a specificity of 0.89, what are the most common causes of false positive notifications, and how might these impact clinical workflow and radiologist burden?; How does the performance (sensitivity, specificity, processing time) of Rapid PETN vary across different CT scanner models within the validated brands (GE, Siemens, Toshiba), particularly with older versus newer generation scanners or different reconstruction kernels?; What is the real-world impact on patient outcomes or time-to-treatment when Rapid PETN is integrated into a busy clinical setting compared to standard triage protocols, especially in scenarios with high caseloads?