BriefCase

The BriefCase device is an artificial intelligence-driven software designed to expedite radiological triage by analyzing head CTA images for suspected Large Vessel Occlusion, aiming to prioritize radiologist review through timely notifications. Its methodology leverages cloud-based deep learning for image processing and integrates with hospital workflows via a desktop application, providing non-diagnostic previews to facilitate faster attention to potentially critical stroke cases, supported by clinical data demonstrating high sensitivity and specificity.

BriefCase is a radiological computer-aided triage and notification software developed by Aidoc Medical, Ltd., specifically designed to analyze head CTA images for the detection of suspected Large Vessel Occlusion (LVO) pathologies. Its primary purpose is to assist hospital networks and trained radiologists in optimizing workflow triage by flagging and communicating these potential critical findings. The system employs an artificial intelligence algorithm, based on deep learning, to process DICOM images. The operational methodology involves three modules: the Aidoc Hospital Server for image acquisition, the Aidoc Cloud Server for image processing (including de-identification), and the Aidoc Worklist Application, which provides notifications and a user interface on the radiologist's desktop. Notifications for suspected LVO cases include compressed, non-diagnostic preview images and are delivered in parallel to the standard image interpretation workflow, without altering the original medical images or removing cases from the standard queue. The clinical performance of BriefCase for LVO triage was evaluated in a retrospective, blinded, multicenter study involving 383 cases from three U.S. clinical sites. The software demonstrated a sensitivity of 88.8% and a specificity of 87.2%, surpassing the predefined 80% performance goal for its primary endpoints. Secondary performance metrics included a positive likelihood ratio of 6.9 and a negative likelihood ratio of 0.13, with positive and negative predictive values varying by LVO prevalence (e.g., PPV of 69.8% and NPV of 95.9% at 25% prevalence). A key operational result was the mean time-to-notification of 3.8 minutes for true positive LVO cases, suggesting its potential to facilitate earlier radiologist assessment and triage by prompting preemptive evaluation of critical findings compared to standard chronological review.

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 BriefCase algorithm's performance on LVO detection vary with different CTA scanner manufacturers, resolutions, or contrast media protocols beyond the specified inclusion criteria (e.g., slice thickness 0.5mm-1.0mm)?; What are the long-term maintenance and update strategies for the AI algorithm to ensure sustained accuracy as imaging technologies evolve and new LVO presentation patterns are identified?; What is the observed false positive notification rate in diverse, real-world clinical settings, and what is its potential impact on radiologist workload and the risk of 'alert fatigue'?; How does the 'notification filter' feature (by pathology or center) interact with the core algorithm's sensitivity, and what safeguards exist to prevent inadvertent filtering of true positive LVO cases, especially in complex multi-site hospital networks?; Beyond accelerating radiologist notification, is there data or are studies planned to quantify the downstream impact of BriefCase implementation on patient-centric outcomes, such as time-to-treatment or length of hospital stay for LVO patients?