Detecting an embolic stroke quickly and accurately is a challenge that requires an experienced radiologist to analyze CT scans. Since time is of the essence in these situations, an automated computer system that can identify stroke can help patients get the right treatment before more damage is done.
Researchers at Hong Kong Polytechnic University have been working on just such a system that takes CT scan data as input and puts visual markers where it suspects a thrombus is blocking a vessel. The researchers are working on developing a learning algorithm that would improve the accuracy of the system by studying the accurate diagnoses from other scans. From Dr Fuk-hay Tang of Hong Kong PolyU:
“The clock is ticking for stroke patients. Medications taken in three hours from the onset of stroke are deemed most effective. Chances of recovery decrease with every minute passing by. It usually takes half an hour for the ambulance to arrive at the hospital, at best. Then, another 45 minutes to 1 hour are needed for CT or MRI scans after the patient has been checked and dispatched for the test, which means some waiting and time will slip by. Afterwards, the brain scan will take another 10 to 15 minutes. If our tool can help doctors arrive at a diagnosis in 10 minutes, the shorter response time will make meeting the target more achievable.”
“It might come in handy for physicians with less experience in stroke,” added Dr Tang, “and patient care can be maintained in hospitals where human and other vital resources are already stretched to the limit.”
Press release: Beating the clock for sufferers of ischemic stroke…