HealthICH

HealthICH is a radiological computer-aided triage and notification software that employs an artificial intelligence algorithm to analyze adult non-contrast head CT scans for suspected intracranial hemorrhage, aiming to expedite case review through worklist prioritization. Its science lies in AI-driven image analysis providing case-level alerts to PACS/workstations, with performance validated by high sensitivity and specificity in detecting ICH, thus facilitating faster clinical attention without making direct diagnostic claims.

HealthICH, by Zebra Medical Vision Ltd., is an artificial intelligence-driven software tool designed to aid in the clinical assessment of adult non-contrast head CT scans by identifying features suggestive of intracranial hemorrhage. Its purpose is to facilitate worklist prioritization and triage within a medical care environment by analyzing CT cases and notifying the PACS/workstation of suspected critical findings. HealthICH operates in parallel to standard workflows, does not direct user attention to specific image locations, and is not intended for standalone diagnostic decision-making or to rule out hemorrhage; its output is a case-level notification for clinicians to review the prioritized scan. The device utilizes an AI algorithm to analyze validated head CT studies, integrating with existing systems via the Zebra Imaging Analytics platform without its own graphical user interface. Performance was validated in a retrospective study of 427 anonymized head CT cases (199 positive for ICH, 228 negative), truthed by US Board Certified neuro-radiologists. This validation showed a sensitivity of 94.47% and specificity of 92.54%, with an overall agreement of 93% against the ground truth. The average per-case processing time was 48.67 seconds, aiming to reduce the time to notify physicians about potential intracranial hemorrhages.

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 HealthICH algorithm's performance vary across different CT scanner manufacturers, acquisition parameters, and reconstruction kernels?; What was the spectrum of intracranial hemorrhage subtypes (e.g., subdural, epidural, intraparenchymal, subarachnoid) and volumes present in the validation dataset, and did performance differ significantly among them?; Beyond worklist prioritization, are there studies planned or underway to evaluate the impact of HealthICH on downstream clinical outcomes, such as time-to-intervention or length of hospital stay for patients with intracranial hemorrhage?; How does the system differentiate between true intracranial hemorrhage and potential mimics (e.g., calcifications, beam hardening artifacts, contrast enhancement if inadvertently included), and what is the typical false positive notification rate in real-world clinical deployment?; What mechanisms are in place for ongoing monitoring and retraining of the HealthICH algorithm to ensure sustained performance as imaging technology evolves or new pathological presentations emerge?