A new study presented at the Society of NeuroInterventional Surgery’s (SNIS) 19th annual meeting (25–29 July, Toronto, Canada) shows that artificial intelligence (AI) technology can identify when a patient is having a stroke caused by emergent large vessel occlusion (LVO)—therefore making them a candidate for endovascular therapy (EVT).
Getting a diagnosis quickly is critical and can be the difference between a life of disability versus rehabilitation for stroke patients, an SNIS press release notes.
The aforementioned study, “AI-based gaze deviation detection to aid LVO diagnosis in NCCT”, used AI algorithms to detect gaze deviation from a non-contrast computed tomography (NCCT) scan. These scans predict if a patient is having an ischaemic stroke caused by an LVO.
If a patient is having this type of stroke, they can receive EVT to treat it. EVT, also known as thrombectomy, is a minimally invasive procedure that uses catheters to reopen blocked arteries in the brain quickly, potentially reducing or preventing long-term disability or damage to the patient’s brain.
“Endovascular thrombectomy for acute ischaemic stroke is one of the most effective therapies in medicine and its efficacy is critically dependent on time,” said lead author of the study Jason Tarpley (Pacific Stroke and Aneurysm Center, Santa Monica, USA). “Simply put, the faster we act, the better our stroke patients’ outcomes will be. Our results represent an advance that has the potential to speed up the identification of LVO stroke during the triage process at the hospital.”
In the study, the AI algorithm was trained using a set of 200 scans to identify gaze direction. It then identified clinical symptoms of ipsiversive gaze deviation in 116 stroke patients with LVO, treated with EVT. The deviation was calculated by measuring the angle between the gaze direction and the midline of the brain.
The algorithm correctly identified 79% of proximal occlusions with an ipsiversive gaze deviation, but had a more difficult time identifying cases with less severe symptoms.