Artificial Intelligence – Deep learning algorithm to detect glaucoma

Why this research is important:

Fourth-year medical student Waldir Rodrigues De Souza Jr from the University of Otago, was the Gordon Sanderson scholar 2019. His research into applying artificial intelligence (AI) to detect glaucoma was presented at the RANZCO conference.

De Souza said, “using artificial intelligence to detect glaucoma is amazing and possible, but it is just the beginning.”

In line with worldwide studies, De Souza’s research focused on training a deep learning algorithm to detect glaucoma, which the team achieved with “exciting sensitivity and specificity,” he said. “We used three image modalities to train our convoluted neural network: colour disc photos, visual fields, and optical coherence tomography (OCT) scan.

Key Research Findings:

AI performed the best using OCTs. With a single central slice of the optic nerve head, the algorithm was able to correctly detect eyes with glaucoma 92.6% of the time and eyes without glaucoma 95.2% of the time.”