FDA Approves Marketing for First AI Device for Diabetic Retinopathy Detection

The US Food and Drug Administration (FDA) approved the marketing of the first medical device that uses artificial intelligence (AI) to detect notable eye disease diabetic retinopathy (DR) in adults with diabetes.

The device, IDx-DR, is a software program that uses an AI algorithm to analyze images taken with retinal camera Topcon NW400. Physicians can then upload the digital images of patient’s retinas to an IDx-DR applicable cloud server. Sufficient-quality images will provide physicians results that either detect DR or report negative results. Patients with positive results will be then referred to an eye care provider for further evaluation.

The device from privately-held AI diagnostics company IDx makes FDA history as the first of its kind authorized for marketing that provides a screening decision without need of clinician interpretation — allowing health care providers outside of an eye care specialty to use it on patients.

DR, the most common cause of vision loss among 30 million-plus Americans with diabetes, occurs when high levels of blood sugar lead to eventual damage in retina blood vessels. The condition's growing prevalence and the new therapies that are emerging to counter it, have resulted in projected compound annual growth rate of 6.8% in the global DR market. Experts expect its value to increase from $5,934.1 million in 2014 to $10.11 billion in 2022.

IDx-DR is the first device authorized for marketing that provides a screening decision without the need for a clinician to also interpret the image or results, which makes it usable by health care providers who may not normally be involved in eye care. It was approved on the backing of clinical study results of retinal images obtained from 900 patients with diabetes. The device was able to accurately identify the presence of more-than-mild DR 87.4% of the time, and was able to accurately identify patients who did not have more-than-mild DR 89.5% of the time.

Michael D. Abràmoff, MD, PhD, IDx founder and president, told EyeNet in a November 2017 interview that prior to the IDx-DR software, researchers had been using machine learning to combine outputs of various detectors into diagnostic outputs.

“We wanted to see if—instead of using mathematical equations as detectors—we could use small convolutional neural networks to find these lesions,” Abràmoff said. “And that led to a statistically significant improvement in performance.”

He credited the accurate trial results to the software’s exposure to reliable reference standards.

“You need to start with datasets that everyone agrees are validated,” Abràmoff said. “You cannot just take any set from a retinal clinic and say, well, here’s a set of bad disease and here’s a group of normal.”

The FDA reviewed IDx-DR under the De Novo premarket review pathway, which is designated for low-to moderate-risk novel devices seeking regulation for market with no prior legal devices. It was also previously granted Breakthrough Device designation by the FDA.

Malvina Eydelman, MD, director of the Division of Ophthalmic, and Ear, Nose and Throat Devices at the FDA's Center for Devices and Radiological Health, said in a statement that many patients with diabetes are not adequately screened for DR, due to approximately 50% of them not seeing an eye doctor annually. As a result, there’s a heightened need for early retinopathy detection.

“Today's decision permits the marketing of a novel artificial intelligence technology that can be used in a primary care doctor's office. The FDA will continue to facilitate the availability of safe and effective digital health devices that may improve patient access to needed health care,” Eydelman said.