[This article by Raynard Churchwell, with photo of Mahmut Karakaya and Ramazan Aygun by Darnell Wilburn, first appeared on the Kennesaw State University website, republished with permission]

Kennesaw State University researchers Mahmut Karakaya and Ramazan Aygun recently earned a National Institutes of Health (NIH) grant to support their technology-driven approach to addressing diabetic eye disease.

The project will focus on diabetic retinopathy, a complication of diabetes that damages the tiny blood vessels in the retina. Left undetected, diabetic retinopathy can lead to vision loss without early symptoms. According to the Centers for Disease Control and Prevention, more than 9 million people in the U.S. suffer from the serious eye condition.

Supported by the NIH grant, Karakaya and Aygun will develop a smartphone-based retinal imaging system to help detect the condition earlier and more affordably through the use of artificial intelligence. Karakaya began exploring the technology a decade ago.

“We were experimenting with smartphones to scan invisible barcodes,” said Karakaya, an assistant professor of computer science in KSU’s College of Computing and Software Engineering. “That caused me to think, ‘What if we used smartphones to capture medical images?’”

Later, the research pivoted to retinal imaging, recognizing that existing smartphone tools could capture images but not analyze them. Combining smartphone hardware with machine learning, the duo aimed to create a diagnostic tool that was both accessible and accurate.

Egypt was selected as the testing location because it has a diabetes prevalence of over 20 percent. Through an Egypt-based collaborator, Karakaya and Aygun formed partnerships with local medical professionals to support the project. While Egypt is the starting point, the technology is intended for global use, including rural areas in the U.S.

Unlike traditional equipment costing as much as $50,000, their system relies on a smartphone, a lens attachment, and AI to screen for diabetic retinopathy. A patient’s retinal image is captured, assessed for quality, and analyzed by AI. If abnormalities are found, the system recommends a follow-up with a specialist.

“This approach is game-changing for primary care,” said Aygun, an associate professor of computer science. “With this tool, we can reach people where they are without expensive equipment or needing a specialist on site.”

The team is also developing WisdomNet, an AI framework designed to recognize its own uncertainty and defer complex cases to human doctors.

“The goal isn’t just to build an accurate model, but a trustworthy one,” Aygun added.

To make the AI more robust, they’re incorporating expert knowledge and plan to use gaze-tracking to teach the algorithm how doctors evaluate retinal images. Their aim is to reach at least 80% diagnostic accuracy by the end of the project.

Students at all levels, from undergraduates to Ph.D. candidates, are contributing to the work.

“They are gaining hands-on research experience that will shape their future careers,” Karakaya said. “They’re not just learning how to code or train models. They’re learning how to build ethical, reliable tools that impact lives.”

Long-term, the faculty members hope the system will be scalable enough for use in mobile clinics, pop-up screenings, and even remote monitoring by community health workers. With proper support and deployment, the technology could address broader healthcare gaps beyond diabetic retinopathy, eventually adapting to detect other vision-threatening diseases like glaucoma and macular degeneration.

“Accessibility is at the core of what we’re doing,” said Karakaya. “This is about removing barriers cost, distance, technology and giving people the power to act early.”