AI-assisted triage tool reduces time to diagnose stroke
A new stroke triage tool powered by AI is now available in the UK. AUTOStroke from Canon Medical analyses and categorises diagnostic brain images automatically following a CT scan to detect signs of ischaemic and haemorrhagic stroke in 30 seconds.
The imaging tool integrates a comprehensive set of stroke applications including: Non-contrast CT Intracranial Haemorrhage that helps quickly locate multiple variations of intracranial haemorrhage when time is critical; Non-contrast CT ASPECTS to deliver scoring of early ischaemic stroke signs that are routinely challenging to detect; CT Large Vessel Occlusion to help locate occlusions in cerebral infarction patients; and CT Perfusion Maps that are created using validated Bayesian CTP+ algorithms to help indicate potential regions of penumbra and infarct core.
The tool automates diagnosis with zero clicks from the CT scan to a clinical decision, consolidating scan results into a single summary and alerting clinicians to any abnormalities.
“While we have approximately 90 minutes to treat a patient with stroke; up to 30 minutes of that time might be spent on imaging alone,” said assistant professor in residence for radiological sciences and computer science Dr Peter Chang at University of California Irvine (UCI), who worked closely with Canon Medical Systems on the development of the new tool. “Using AI for strokes enables rapid interpretation within half a minute from when images are acquired, facilitating rapid and robust treatment and turnaround times that would be very difficult without that type of system. AUTOStroke is used at UCI to automatically triage and evaluate every patient with suspected stroke who comes into the emergency room.”
Picture: AUTOStroke integrates stroke applications including (left) non-contrast CT ASPECTS that delivers scoring of early ischaemic stroke signs and (right) CT Large Vessel Occlusion to help locate occlusions in cerebral infarction patients.
Published on page 9 of the May 2022 issue of RAD Magazine.
