Harrison.ai Secures the US FDA 510(k) Clearance for Acute Infarct Triage on Non-Contrast CT Brain

Why It Matters

The clearance enables faster, more accurate stroke triage in emergency settings, potentially reducing treatment delays and improving patient outcomes while expanding AI’s role in radiology without additional contrast procedures.

Key Takeaways

• FDA grants 510(k) clearance for Harrison.ai infarct triage.
• Sensitivity ~89% on thin CT slices, 86% on thick slices.
• Outperforms existing LVO AI tools lacking infarct detection.
• Works on non‑contrast CT, eliminating need for contrast agents.
• Sensitivity and specificity exceed 80% across six vascular territories.

Pulse Analysis

Stroke remains a leading cause of disability, and rapid diagnosis is critical for administering time‑sensitive therapies such as thrombolysis or thrombectomy. While non‑contrast CT is the fastest imaging modality available in most emergency departments, its interpretation for early ischemic changes can be subtle and operator‑dependent. Artificial intelligence has emerged as a solution, offering consistent, quantitative assessments that can flag potential infarcts before a radiologist’s read, thereby shortening door‑to‑needle times.

Harrison.ai’s newly cleared Acute Infarct Triage software builds on this premise by delivering high‑sensitivity detection across both thin and thick CT slices. Compared with the only other FDA‑cleared LVO AI device—limited to vessel occlusion detection and achieving 63.5% sensitivity—the new tool reaches nearly 90% sensitivity for infarct identification, while maintaining specificity above 80% across all six major cerebral vascular territories. Importantly, it operates on standard non‑contrast scans, avoiding the need for contrast‑enhanced CTA, which can be contraindicated in patients with renal impairment or allergies. This performance profile positions the solution as a more comprehensive triage aid, capable of guiding clinicians toward both occlusion and tissue‑level decisions.

The clearance signals a broader regulatory acceptance of AI that goes beyond binary vessel detection toward nuanced tissue characterization. Hospitals can now integrate a single AI workflow into existing CT protocols, potentially reducing the need for multiple imaging studies and associated costs. As payers and providers seek efficiency gains, the market for AI‑driven stroke tools is likely to expand, prompting further investment in algorithms that combine speed, accuracy, and ease of deployment. Harrison.ai’s achievement may accelerate competitive development and set new benchmarks for clinical validation in the radiology AI space.