AI for a eye: New tool to take pressure off emergency doctors

In 2009, a 12-year-old boy fell off his bike and hit his head on the pavement in Maryborough in central Victoria. After experiencing headaches, his mother took him to the local hospital where he was diagnosed with pressure on the brain.

The doctor on duty had to act quickly, using a drill from the maintenance room to relieve the pressure and bleeding on the boy’s brain with guidance over the phone from a Melbourne neurosurgeon.

Thanks to the quick help, the boy could then be airlifted to a major children’s hospital and was safely released in time for his 13th birthday.

“That story really sat with me,” said Associate Professor Mojtaba Golzan, an expert in vision science from the UTS School of Clinical and Health Sciences.

"If doctors had been able to detect the pressure on his brain earlier, they would have been able to make arrangements to transfer him to a larger hospital."

It’s an experience Golzan has taken into developing a new technology BENAH.ai that combines retinal imaging with artificial intelligence to identify intercranial pressure quickly and enable triaging for more detailed diagnosis.

“When somebody presents to an emergency department with a headache, clinicians need to quickly rule out whether they've got dangerous pressure building up within the brain,” he said.

“Currently, it can take a couple of hours or a couple of days to rule that out using brain imaging and lumbar punctures. With our technology, we’re trying to get that result in a couple of minutes.”

BENAH.ai is a software-as-a-medical-device that takes images from the back of the eye and puts them through an artificial intelligence platform which analyses them against features linked to abnormal brain pressure.

It then provides a probability score of how likely it is the patient has abnormal brain pressure.

“Swelling of the optic disc or optic nerve head – the medical term for it is papilledema – is associated with an increase in intracranial pressure,” Golzan said.

“But identifying it is not as simple as scanning a patient’s eye. Not all swelling of the optic nerve is caused by intracranial pressure, and not all pressure leads to a swelling in the eye.”

“It’s very difficult even for medical specialists and many emergency clinicians can struggle to fully understand the complexities of identifying it."

"They often need to bring in ophthalmologists or neurologists, which takes time.”

“That’s where our technology comes in – inspired by the Persian word BENAH (“bee.naa”) meaning “to see”.”

The artificial intelligence driving BENAH.ai is being developed to identify the key visual features in the back of the eye that correlate to increased brain pressure. 

They have drawn on a large dataset of eye imaging and diagnostic data from historical patients from across various clinical sites to build and refine their models.