AI-Powered Brain Scans Dramatically Increase Stroke Thrombectomy Rates in England
Table of Contents
New research published in The Lancet Digital Health demonstrates a notable rise in mechanical thrombectomy procedures for acute stroke patients following the implementation of artificial intelligence (AI) for rapid brain imaging analysis.
A groundbreaking study reveals that the use of AI to analyze brain scans is dramatically improving access to life-saving stroke treatment in England. The research, conducted across multiple hospitals, shows a considerable increase in the number of patients receiving mechanical thrombectomy – a procedure to remove blood clots from the brain – after the introduction of AI-powered imaging software.
The Challenge of Timely Stroke Care
Acute stroke care hinges on swift diagnosis and intervention. Patients are initially assessed at Neurovascular Units (NVUs), where initial assessments and imaging take place.These NVUs often lack the specialized expertise and infrastructure for interventional neuroradiology, which is concentrated in dedicated centers.
The critical bottleneck lies in the speed of identifying stroke etiology and transferring eligible patients to these specialized centers.Delays directly translate to lower rates of thrombectomy. In 2024, the UK’s thrombectomy rate stood at just 3.9%, falling below the European average despite a potential of 15%.
NHS Embraces AI for Faster Diagnosis
To address this challenge, the National Health Service (NHS) began deploying Brainomix 360***, an AI software developed by Brainomix in Oxford, across 26 of England’s 107 stroke centers in march 2020. This software provides real-time interpretation of CT, angio-CT, and perfusion CT scans, aiding in the rapid identification of stroke type and severity.
Researchers conducted a prospective observational study using data from the national stroke registry at King’s College London,spanning five years (January 1,2019 – December 31,2023).The study compared outcomes before and after the widespread implementation of the AI software in 20 local NVUs and six referral centers.
Dramatic Increase in Thrombectomy Rates
The results are compelling. Among 452,952 patients admitted with suspected stroke, 71,017 were diagnosed with ischemic stroke at the 26 evaluation sites. Before the AI software,the thrombectomy rate at these centers was 2.3%. following implementation, this rate doubled to 4.6% – a 100% increase.
even in hospitals without the AI software, a significant increase was observed, rising from 1.6% to 2.6% (a 62.5% increase).
AI Reduces Critical Response Time
Data analysis revealed a statistically significant association between AI implementation and increased thrombectomy rates. The odds ratio (OR) was 1.5 for evaluation sites compared to non-evaluation sites,and 1.7 after software deployment compared to before. Moreover, the software demonstrably reduced the time to transfer patients.The median time between arrival at the local NVU and departure to a referral center decreased from 192 minutes in sites without AI to 128 minutes in sites using the software – a substantial reduction in critical response time.
The use of the software also proved to be a predictive factor for the management of intravenous thrombolysis.
Improved Outcomes, But Further Research Needed
The study also found that AI use was associated with a higher probability of good clinical outcome (OR of 1.16). However, it did not result in reduced in-hospital mortality.
According to the researchers, this study represents the largest prospective evaluation of AI’s impact on clinical care in stroke. The findings strongly support the integration of this technology into standard stroke care pathways.
“The results demonstrate a benefit on access to thrombectomy,” the researchers conclude, “and support recommendations in favor of the integration of this tool into care pathways.”
( The Lancet Digital Health, December 2 post)
