Skip to main content

diabetic retinopathy

Artificial Intelligence Getting Smarter! Innovations from the Vision Field

Posted on by Michael F. Chiang, M.D., National Eye Institute

AI. Photograph of retina

One of many health risks premature infants face is retinopathy of prematurity (ROP), a leading cause of childhood blindness worldwide. ROP causes abnormal blood vessel growth in the light-sensing eye tissue called the retina. Left untreated, ROP can lead to lead to scarring, retinal detachment, and blindness. It’s the disease that caused singer and songwriter Stevie Wonder to lose his vision.

Now, effective treatments are available—if the disease is diagnosed early and accurately. Advancements in neonatal care have led to the survival of extremely premature infants, who are at highest risk for severe ROP. Despite major advancements in diagnosis and treatment, tragically, about 600 infants in the U.S. still go blind each year from ROP. This disease is difficult to diagnose and manage, even for the most experienced ophthalmologists. And the challenges are much worse in remote corners of the world that have limited access to ophthalmic and neonatal care.

Caption: Image of a neonatal retina prior to AI processing. Left: Image of a premature infant retina showing signs of severe ROP with large, twisted blood vessels; Right: Normal neonatal retina by comparison. Credit: Casey Eye Institute, Oregon Health and Science University, Portland, and National Eye Institute, NIH

Artificial intelligence (AI) is helping bridge these gaps. Prior to my tenure as National Eye Institute (NEI) director, I helped develop a system called i-ROP Deep Learning (i-ROP DL), which automates the identification of ROP. In essence, we trained a computer to identify subtle abnormalities in retinal blood vessels from thousands of images of premature infant retinas. Strikingly, the i-ROP DL artificial intelligence system outperformed even international ROP experts [1]. This has enormous potential to improve the quality and delivery of eye care to premature infants worldwide.

Of course, the promise of medical artificial intelligence extends far beyond ROP. In 2018, the FDA approved the first autonomous AI-based diagnostic tool in any field of medicine [2]. Called IDx-DR, the system streamlines screening for diabetic retinopathy (DR), and its results require no interpretation by a doctor. DR occurs when blood vessels in the retina grow irregularly, bleed, and potentially cause blindness. About 34 million people in the U.S. have diabetes, and each is at risk for DR.

As with ROP, early diagnosis and intervention is crucial to preventing vision loss to DR. The American Diabetes Association recommends people with diabetes see an eye care provider annually to have their retinas examined for signs of DR. Yet fewer than 50 percent of Americans with diabetes receive these annual eye exams.

The IDx-DR system was conceived by Michael Abramoff, an ophthalmologist and AI expert at the University of Iowa, Iowa City. With NEI funding, Abramoff used deep learning to design a system for use in a primary-care medical setting. A technician with minimal ophthalmology training can use the IDx-DR system to scan a patient’s retinas and get results indicating whether a patient should be sent to an eye specialist for follow-up evaluation or to return for another scan in 12 months.

Caption: The IDx-DR is the first FDA-approved system for diagnostic screening of diabetic retinopathy. It’s designed to be used in a primary care setting. Results determine whether a patient needs immediate follow-up. Credit: Digital Diagnostics, Coralville, IA.

Many other methodological innovations in AI have occurred in ophthalmology. That’s because imaging is so crucial to disease diagnosis and clinical outcome data are so readily available. As a result, AI-based diagnostic systems are in development for many other eye diseases, including cataract, age-related macular degeneration (AMD), and glaucoma.

Rapid advances in AI are occurring in other medical fields, such as radiology, cardiology, and dermatology. But disease diagnosis is just one of many applications for AI. Neurobiologists are using AI to answer questions about retinal and brain circuitry, disease modeling, microsurgical devices, and drug discovery.

If it sounds too good to be true, it may be. There’s a lot of work that remains to be done. Significant challenges to AI utilization in science and medicine persist. For example, researchers from the University of Washington, Seattle, last year tested seven AI-based screening algorithms that were designed to detect DR. They found under real-world conditions that only one outperformed human screeners [3]. A key problem is these AI algorithms need to be trained with more diverse images and data, including a wider range of races, ethnicities, and populations—as well as different types of cameras.

How do we address these gaps in knowledge? We’ll need larger datasets, a collaborative culture of sharing data and software libraries, broader validation studies, and algorithms to address health inequities and to avoid bias. The NIH Common Fund’s Bridge to Artificial Intelligence (Bridge2AI) project and NIH’s Artificial Intelligence/Machine Learning Consortium to Advance Health Equity and Researcher Diversity (AIM-AHEAD) Program project will be major steps toward addressing those gaps.

So, yes—AI is getting smarter. But harnessing its full power will rely on scientists and clinicians getting smarter, too.

References:

[1] Automated diagnosis of plus disease in retinopathy of prematurity using deep convolutional neural networks. Brown JM, Campbell JP, Beers A, Chang K, Ostmo S, Chan RVP, Dy J, Erdogmus D, Ioannidis S, Kalpathy-Cramer J, Chiang MF; Imaging and Informatics in Retinopathy of Prematurity (i-ROP) Research Consortium. JAMA Ophthalmol. 2018 Jul 1;136(7):803-810.

[2] FDA permits marketing of artificial intelligence-based device to detect certain diabetes-related eye problems. Food and Drug Administration. April 11, 2018.

[3] Multicenter, head-to-head, real-world validation study of seven automated artificial intelligence diabetic retinopathy screening systems. Lee AY, Yanagihara RT, Lee CS, Blazes M, Jung HC, Chee YE, Gencarella MD, Gee H, Maa AY, Cockerham GC, Lynch M, Boyko EJ. Diabetes Care. 2021 May;44(5):1168-1175.

Links:

Retinopathy of Prematurity (National Eye Institute/NIH)

Diabetic Eye Disease (NEI)

NEI Research News

Michael Abramoff (University of Iowa, Iowa City)

Bridge to Artificial Intelligence (Common Fund/NIH)

Artificial Intelligence/Machine Learning Consortium to Advance Health Equity and Researcher Diversity (AIM-AHEAD) Program (NIH)

[Note: Acting NIH Director Lawrence Tabak has asked the heads of NIH’s institutes and centers to contribute occasional guest posts to the blog as a way to highlight some of the cool science that they support and conduct. This is the second in the series of NIH institute and center guest posts that will run until a new permanent NIH director is in place.]


Creative Minds: Reverse Engineering Vision

Posted on by Dr. Francis Collins

Networks of neurons in the mouse retina

Caption: Networks of neurons in the mouse retina. Green cells form a special electrically coupled network; red cells express a distinctive fluorescent marker to distinguish them from other cells; blue cells are tagged with an antibody against an enzyme that makes nitric oxide, important in retinal signaling. Such images help to identify retinal cell types, their signaling molecules, and their patterns of connectivity.
Credit: Jason Jacoby and Gregory Schwartz, Northwestern University

For Gregory Schwartz, working in total darkness has its benefits. Only in the pitch black can Schwartz isolate resting neurons from the eye’s retina and stimulate them with their natural input—light—to get them to fire electrical signals. Such signals not only provide a readout of the intrinsic properties of each neuron, but information that enables the vision researcher to deduce how it functions and forges connections with other neurons.

The retina is the light-sensitive neural tissue that lines the back of the eye. Although only about the size of a postage stamp, each of our retinas contains an estimated 130 million cells and more than 100 distinct cell types. These cells are organized into multiple information-processing layers that work together to absorb light and translate it into electrical signals that stream via the optic nerve to the appropriate visual center in the brain. Like other parts of the eye, the retina can break down, and retinal diseases, including age-related macular degeneration, retinitis pigmentosa, and diabetic retinopathy, continue to be leading causes of vision loss and blindness worldwide.

In his lab at Northwestern University’s Feinberg School of Medicine, Chicago, Schwartz performs basic research that is part of a much larger effort among vision researchers to assemble a parts list that accounts for all of the cell types needed to make a retina. Once Schwartz and others get closer to wrapping up this list, the next step will be to work out the details of the internal wiring of the retina to understand better how it generates visual signals. It’s the kind of information that holds the key for detecting retinal diseases earlier and more precisely, fixing miswired circuits that affect vision, and perhaps even one day creating an improved prosthetic retina.


Big Data Study Reveals Possible Subtypes of Type 2 Diabetes

Posted on by Dr. Francis Collins

Computational model

Caption: Computational model showing study participants with type 2 diabetes grouped into three subtypes, based on similarities in data contained in their electronic health records. Such information included age, gender (red/orange/yellow indicates females; blue/green, males), health history, and a range of routine laboratory and medical tests.
Credit: Dudley Lab, Icahn School of Medicine at Mount Sinai, New York

In recent years, there’s been a lot of talk about how “Big Data” stands to revolutionize biomedical research. Indeed, we’ve already gained many new insights into health and disease thanks to the power of new technologies to generate astonishing amounts of molecular data—DNA sequences, epigenetic marks, and metabolic signatures, to name a few. But what’s often overlooked is the value of combining all that with a more mundane type of Big Data: the vast trove of clinical information contained in electronic health records (EHRs).

In a recent study in Science Translational Medicine  [1], NIH-funded researchers demonstrated the tremendous potential of using EHRs, combined with genome-wide analysis, to learn more about a common, chronic disease—type 2 diabetes. Sifting through the EHR and genomic data of more than 11,000 volunteers, the researchers uncovered what appear to be three distinct subtypes of type 2 diabetes. Not only does this work have implications for efforts to reduce this leading cause of death and disability, it provides a sneak peek at the kind of discoveries that will be made possible by the new Precision Medicine Initiative’s national research cohort, which will enroll 1 million or more volunteers who agree to share their EHRs and genomic information.