AI and glaucoma: "With our technique, you’d be able to quantify the earliest change"

Researchers have combined OCT technology, adaptive optics and deep neural networks to aid the detection of neurodegenerative disease


Scientists at Duke University in the US have utilised optical coherence tomography (OCT), adaptive optics and deep neural networks to assist the detection and management of neurodegenerative conditions such as glaucoma.

The research, which was published in Optica, involved developing a deep learning algorithm that can identify and trace the shapes of ganglion cells from adaptive optics OCT scans.

Testing the algorithm on both scans from healthy and glaucomatous eyes, the framework was able to accurately segment ganglion cells from both samples and also differentiate whether scans were from healthy patients or those with glaucoma.

Post doctoral researcher at Duke University, Somayyeh Soltanian-Zadeh, explained that the technology performed better than human experts in the task.

“It’s superior to other state-of-the-art networks that can process volumetric biomedical images,” she said.

Professor of Biomedical Engineering at Duke University, Sina Farsiu, highlighted that the technology could potentially shorten clinical trials because clinicians would be able to see and measure initial effects of treatment.

“With our technique, you'd be able to quantify the earliest change,” he said.