The role of artificial intelligence in eye care
We are no strangers to artificial intelligence in 2021. With technology like face recognition, personal voice assistants, and perfectly curated digital ads, artificial intelligence (AI) is part of our daily lives. But did you know that AI has also made its way into healthcare, including eye care?
What is artificial intelligence?
Artificial intelligence is a category of computer technology that is meant to mimic the intelligence or behavioural pattern of humans. It is a very complex system, made up of two subsets of systems. Firstly, machine learning is a technique that describes how computers can learn from data without a complex set of rules. Secondly, deep learning is a way to perform machine learning that is inspired by our brain’s own network of neurons.
How is artificial intelligence being used in eye care?
AI has already been in use in healthcare, specifically for the analysis of medical imaging. For example, AI helps detect tuberculosis from chest x-rays, and certain skin and lung cancers.
Ophthalmologists already use very sophisticated technology to examine, diagnose, and treat various eye conditions but AI offers an exciting advance to the field. Deep learning has been applied to fundus photographs (photos of the back of the eye, including the retina), optical coherence tomography and visual fields in order to detect diabetic retinopathy, age-related macular degeneration (AMD), and glaucoma, all of which are the major global causes of blindness.
There are several examples of AI being used in eye care, such as:
- DeepMind, a British AI company acquired by Google, is working with Moorfields Eye Hospital in London to predict the development of exudative AMD, a serious form of the disease.
- An AI algorithm developed by researchers at New York Eye and Ear Infirmary of Mount Sinai to rapidly and accurately detect AMD. It also offers an inexpensive solution to telemedicine, where cameras could be set up in kiosks to examine patients’ eyes. The advantage is that eye exams do not need to be delayed and screenings for the serious condition can continue, contributing to saved vision for many.
- In 2018, the first autonomous diagnosis system was approved for detecting diabetic retinopathy in the United States. Created by retinal specialist Michael Abràmoff, the AI program, IDx-DR, was able to identify more-than-mild diabetic retinopathy about 87% of the time and correctly identified people without the condition about 90% of the time.
Overall, AI’s strength in the field of eye care is its ability to analyze images. Eye exams usually include taking high-quality images and maps of the eye’s different parts. If an AI algorithm is taught to recognize disease in these types of images, it can improve the speed and accuracy of large-scale screening programs. It can also improve access to such exams by potentially allowing for remote imaging.
In a study published in the British Journal of Ophthalmology, 30,000 patient scans (120,000 images) in the English Diabetic Eye Screening Programme (DESP) were used to look for signs of damage from diabetic retinopathy. The technology was extremely accurate at detecting damage that required a referral to a specialist (95.7%) and detecting moderate to severe retinopathy that could lead to vision loss (100%).
This is extremely helpful, given that by the year 2040, an estimated 600 million people will have diabetes, with a third of those having diabetic retinopathy. Since only about half of diabetics have their eyes examined yearly as suggested, technology like this could help cut down wait times, predict prognosis, and reduce unnecessary vision loss or blindness.
Challenges of artificial intelligence
Artificial intelligence does have certain limitations. It learns from what it is given, so it needs large enough data sets (e.g. images) to understand the complexity of a given situation (e.g. what a particular eye disease looks like). Without enough data to learn from, AI could miss something.
The other major challenge of artificial intelligence, especially in healthcare settings, is that it must gain the trust of both patients and physicians. A term often reference with AI is the “black box problem”; this describes how we don’t really know how or why an algorithm makes its decision, we can only observe the inputs and outputs but not the internal workings.
Will AI replace my eye doctor?
No, but it can help them. AI only looks for what you tell it to look for, so using it during routine eye exams, for example, would still be a collaborative process with your ophthalmologist. However, it could save significant screening time and allow doctors to spend more time giving personalized care. Rapid diagnosis with AI also means that you can treat more people. Lastly, AI opens up the possibility for remote diagnosis in certain circumstances, which can help in cases where wait times are long, or a patient is unable to physically get to a doctor’s office.
The field of artificial intelligence in eye care is very exciting. We already know about the amazing technological advancements that allow us to do laser vision correction procedures, which has been around for more than 30 years and allowed surgeons to perform more than 40 million LASIK surgeries. In fact, LASIK MD surgeons alone have performed over 1 million of those, a track record we’re proud of.
If you’re interested in learning more about the procedures we offer, book your free, no-obligation consultation today!