CINA

CINA is a software tool that utilizes artificial intelligence to analyze non-enhanced head CT scans for intracranial hemorrhage and CT angiographies for large vessel occlusions, aiming to streamline radiological workflow triage. Its methodology involves algorithmic image processing on a standalone platform that provides non-diagnostic notifications and preview images to clinicians, thereby enabling faster prioritization of potentially critical cases without altering the standard diagnostic pathway.

CINA is a radiological computer-aided triage and notification software designed for analyzing non-enhanced head CT images for suspected Intracranial Hemorrhage (ICH) and CT angiographies of the head for potential large vessel occlusion (LVO). Its purpose is to assist hospital networks and trained radiologists in optimizing workflow triage by flagging and communicating these suspected positive findings. The system employs an artificial intelligence algorithm to process DICOM images, running dedicated applications for ICH and LVO detection. CINA operates on standard server/workstation hardware and integrates with existing hospital radiology infrastructures, presenting notifications via a standalone web application. These notifications, which include active pop-ups and passive worklist icons, are accompanied by compressed, non-diagnostic preview images and are delivered in parallel to the standard image interpretation process, without altering original medical images or disrupting the standard worklist queue. The performance of CINA was evaluated in a retrospective, multicenter study involving 814 cases for ICH detection and 476 cases for LVO detection. For ICH, the software demonstrated a sensitivity of 91.4% and a specificity of 97.5%, with an Area Under the Curve (AUC) of 0.94. For LVO, the sensitivity was 97.9% and specificity was 97.6%, with an AUC of 0.98. The mean time-to-notification was approximately 21.6 seconds for ICH cases and 34.7 seconds for LVO cases. The software's positive predictive value (PPV) and negative predictive value (NPV) were also assessed across varying disease prevalence rates, demonstrating its utility in clinical settings for prioritizing medical image 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 CINA algorithm differentiate between various subtypes of intracranial hemorrhage, and does its performance vary across these subtypes?; What is the impact of image quality variability (e.g., due to different CT scanner models or protocols) on the CINA software's detection accuracy for ICH and LVO?; Beyond sensitivity and specificity, what is the real-world impact on patient outcomes (e.g., time to treatment, length of hospital stay) when CINA is integrated into a clinical workflow?; How does the CINA platform handle edge cases or rare presentations of ICH and LVO, and what mechanisms are in place for continuous learning or algorithm updates?; What are the computational requirements for deploying CINA, and how does its performance scale with increasing image volume in a large hospital network?