Brainomix 360 Triage Stroke

Brainomix 360 Triage Stroke is a software-only medical device that uses artificial intelligence to analyze non-contrast head CT images and automatically flag suspected large vessel occlusions (LVO) of the intracranial ICA and M1 arteries or intracranial hemorrhage (ICH). It notifies clinicians via web interface or mobile app to assist with workflow triage and prioritization of acute stroke cases within 24 hours of symptom onset, operating in parallel to standard clinical care without replacing angiography or other diagnostic workup.

The subject device differs from the predicate in three ways: (1) an improved ICH detection algorithm using a different deep learning framework and CNN architecture that increased sensitivity for subarachnoid hemorrhage and subdural hematoma subtypes; (2) extended indications to include SAH detection after algorithm improvements; and (3) three configurable modules (Triage ICH, NCCT LVO, and Triage Stroke) selectable within a single device, versus the predicate's fixed Triage Stroke configuration. All configurations use AI/machine learning neural networks operating on NCCT data with notification-only output and no image alteration.

ISO 14971:2019 for risk management; IEC 62304:2015 for software development process; IEC 81001-5-1 for cybersecurity; FDA Cybersecurity Guidance; 21 CFR Part 820.30 design control requirements.

Both devices have substantially equivalent indications for use (triage and notification of suspected LVO or ICH on NCCT), technological approaches (AI/ML neural networks), and clinical roles (parallel workflow prioritization without replacing diagnostic workup). The improved ICH algorithm, SAH extension, and configurable workflows do not raise different safety or effectiveness questions because: (1) the algorithm improvements were clinically validated and showed enhanced performance; (2) SAH inclusion demonstrates better detection rather than introducing new risks; and (3) the three modules employ previously cleared workflows (Triage ICH from K231195, Triage Stroke from K232496, and NCCT LVO using the same Triage Stroke framework) with identical mitigation strategies for mobile preview image limitations.