BriefCase

The BriefCase software employs a deep-learning artificial intelligence algorithm to analyze head CTA images for suspected large and medium vessel occlusions, aiming to expedite clinical workflow triage for adults. It functions as a parallel notification system, alerting specialists to potentially critical cases for prioritized review using non-diagnostic preview images, thereby facilitating earlier attention to time-sensitive conditions without altering original images or providing a diagnosis.

The "BriefCase" software is a radiological computer-aided triage and notification system designed to analyze head CTA images for adults aged 18 years and older. Its primary purpose is to assist hospital networks and appropriately trained medical specialists in prioritizing their workflow by flagging and communicating suspected positive findings of various complete Large Vessel Occlusions (LVOs) – including MCA-M1, PCA-P1, ACA-A1, ICA, Basilar – and Medium Vessel Occlusions (MeVOs) such as MCA-M2, MCA-proximal M3, PCA-P2, PCA-proximal P3, ACA-A2, ACA-proximal A3, and Vertebral-V4. The system utilizes an artificial intelligence algorithm to process DICOM images, which are acquired via an Aidoc Hospital Server, processed on an Aidoc Cloud Server, and then integrated into the workflow through an Aidoc Desktop Application. This application presents notifications of suspected cases, including a compressed, non-diagnostic preview image, operating in parallel with the standard image interpretation process without altering original images or acting as a diagnostic tool itself. The core science of BriefCase involves a deep-learning AI algorithm trained to detect vessel occlusions in head CTA scans. Its efficacy was evaluated in a retrospective, blinded, multicenter study involving 342 cases from five U.S. clinical sites, comparing software performance against a ground truth established by three board-certified neurologists using majority voting. The study demonstrated a sensitivity of 91.3% (95% CI: 83.6%, 96.2%) and a specificity of 85.6% (95% CI: 80.6%, 89.7%), exceeding the 80% performance goal for both metrics. Secondary endpoints showed a mean time-to-notification of 2.23 minutes for true positive cases. The software also reported a Positive Predictive Value (PPV) of 41.3% (95% CI: 34.1%-49.0%) and a Negative Predictive Value (NPV) of 98.9% (95% CI: 4.66%-99.4%). An additional operating point was validated to maximize sensitivity while maintaining a lower bound 95% confidence interval of 80% for specificity. These results suggest the device can effectively flag potential vessel occlusions for earlier review, facilitating preemptive triage.

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 algorithm's performance vary across different types of vessel occlusions (e.g., LVO vs. MeVO, specific artery locations) and in the presence of common imaging artifacts or co-existing pathologies?; What was the inter-rater reliability among the three expert neurologists who established the ground truth in the pivotal study, and how might variations impact the reported sensitivity and specificity?; Given the Positive Predictive Value (PPV) of 41.3%, what is the anticipated impact on clinician workload due to false positive notifications, and are there strategies to mitigate this?; How does the performance of BriefCase change with different CT scanner manufacturers, imaging protocols (beyond slice thickness), or contrast administration techniques encountered in real-world clinical settings?; What is the learning curve for clinicians to effectively integrate BriefCase notifications into their existing triage workflow, and what factors influence its adoption and optimal use in diverse hospital environments?