Massive amounts of data and growing computing power are fueling these advanced, algorithm-based technologies.
Several studies show the potential for AI to help doctors detect eye disease. But more research is needed to prove the technologies do what they set out to. It will take some time for ophthalmologists to trust and use AI-based tools in their practices.
Artificial Intelligence Excels at Image Recognition
AI has been growing in popularity where image analysis is essential to disease diagnosis and treatment. Specialties like radiology, pathology, dermatology and ophthalmology are leading AI research.
AI-based systems — sometimes referred to as deep-learning computers — are trained with many pictures of the eye. The algorithms learn the difference between normal image and abnormal images. Researchers already have tested AI-based systems that use:
- images of the retina to recognize patients at risk for cardiovascular disease, and
- x-ray images to help identify pediatric pneumonia.
Artificial intelligence may also be able to help detect diabetic retinopathy and macular degeneration. It’s estimated that 61 million adults in the United States are at high risk for vision loss. Only half of them visited an eye doctor in the past 12 months. New technologies could get more people to an ophthalmologist sooner for diagnosis and treatment, if they make initial screenings more convenient and accessible. And early treatment can prevent or minimize vision loss. Where medical care is hard to find, AI-based systems can help people who otherwise go without.
“AI won’t replace physicians, but it will make them even more efficient,” says Rahul N. Khurana, MD, an ophthalmologist. “It can help us find and see more of the patients who may be falling through the cracks of the healthcare system.”
Cameras to Detect Diabetic Retinopathy
More than 29 million Americans have diabetes. People with diabetes are at risk for diabetic retinopathy — a potentially blinding eye disease. But people don’t often notice changes in their vision in the disease’s early stages. Worldwide, there are not enough ophthalmologists to take care of all the people who need diabetic eye screenings
That’s where technology like the IDx-DR can help. The IDx-DR is the first FDA-approved AI-based device for detecting diabetic retinopathy. Primary care physicians and other healthcare providers can use it. IDx-DR analyzes images of the eye taken with a retinal camera. The software tells the doctor if a patient should see an ophthalmologist for possible treatment.
And, a March 2018 study published in Ophthalmology, a journal of the American Academy of Ophthalmology, found that disease-detecting AI software can accurately identify early warning signs of diabetic retinopathy. This research provides a strong foundation for future studies.
Software to Recognize Macular Degeneration
Macular degeneration causes central vision loss and is often not noticeable until vision is very blurry. The condition affects more than 10 million Americans. A February 2018 study in Cell Magazineshowed that artificial intelligence-based software could recognize early signs of macular degeneration. To do this, the software was taught using images selected by experts in image assessment. Focusing on these images that give clear diagnostic examples allowed the software to learn from a smaller number of images and produce results faster. This technology may be able to assist physicians in getting more people into treatment sooner.
AI also could help ophthalmologists map and measure how a person’s disease is progressing. Computer systems can rapidly process multiple eye scans and sets of information. AI may one day aid eye doctors in treatment decisions for conditions like macular degeneration, glaucoma and more.