BriefCase

The BriefCase software employs an artificial intelligence deep-learning algorithm to analyze adult head CTA images for M1 Large Vessel Occlusion, intending to expedite radiological triage. By delivering rapid notifications of suspected positive findings to a standalone desktop application, the system aims to assist medical specialists in prioritizing critical cases in parallel with standard image interpretation workflows.

BriefCase is a radiological computer-aided triage and notification software designed to analyze head CTA images in adults (18 years and older) for M1 Large Vessel Occlusion (M1-LVO) pathologies. The system utilizes an artificial intelligence deep-learning algorithm and comprises three modules: the Aidoc Hospital Server for image acquisition, the Aidoc Cloud Server for image processing, and the Aidoc Desktop Application for workflow integration and user notifications. DICOM images are received, filtered based on metadata, de-identified, and then processed chronologically. When the algorithm detects a suspected M1-LVO, it sends a notification to the desktop application, which includes a compressed, non-diagnostic preview image, to assist medical specialists in prioritizing their review of original images within the standard care pathway. A retrospective, blinded, multicenter study evaluated the performance of BriefCase for M1-LVO detection using datasets distinct from those used for algorithm training. The primary endpoints achieved were a sensitivity of 88.8% and a specificity of 87.2%, both exceeding an 80% performance goal. The mean time-to-notification for true positive cases was 3.8 minutes. Other reported metrics include a Negative Predictive Value (NPV) of 99.5% and a Positive Predictive Value (PPV) of 20.1%. The study involved images from patients with a mean age of 65.1 years, acquired using CT scanners from various manufacturers. The device operates in parallel with standard care, not altering original images or the worklist order, and aims to facilitate earlier attention to critical cases.

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 (sensitivity, specificity, PPV, NPV) vary across different CT scanner manufacturers and imaging protocols beyond the stated inclusion criteria for M1-LVO detection?; What is the potential impact of the 20.1% Positive Predictive Value on radiologist workflow, specifically concerning 'alert fatigue', and are there configurable thresholds or secondary checks to mitigate this?; How does the BriefCase software handle variations in image quality (e.g., motion artifacts, contrast timing) or atypical M1-LVO presentations that might challenge the AI algorithm's detection capabilities?; What are the specific computational resource requirements for the Aidoc Hospital Server and Cloud Server components, and what mechanisms ensure data security and patient privacy during the image de-identification and cloud processing stages?; Beyond time-to-notification, has any data been collected on the direct impact of BriefCase M1-LVO triage on downstream clinical metrics, such as door-to-needle time or patient outcomes, in a prospective real-world setting?