August 2020


Artificial Intelligence
Opportunities for artificial intelligence in glaucoma

by Ellen Stodola Editorial Co-Director

Dr. Habash conducts a telehealth hybrid visit with a glaucoma patient. She is doing a live video slit lamp exam with the tech, so she is seeing the whole exam, reviewing the OCT/IOP, etc., and talking to the patient throughout.
Source: Ranya Habash, MD

“Technology helps us lift the weight, but we’re still the ones doing the work. AI and telemedicine won’t replace us, they will augment us.”
—Ranya Habash, MD

With increasing applications for artificial intelligence (AI) in ophthalmology, experts discussed ways that it might be applied to the field of glaucoma.
Valerie Trubnik, MD, said she thinks the retina field is “a little ahead of the game” when it comes to AI. However, there could be uses for it in glaucoma, though Dr. Trubnik noted that “glaucoma makes for a very complicated diagnosis” because there are so many modalities involved.
At this point, Dr. Trubnik said AI is being used more for screening and diagnosis and added that fundus photos and OCT are probably easiest to use, but visual fields can be included as well. With the growing number of glaucoma patients, “we need a better and faster way to sort through these images,” she said.
Ranya Habash, MD, stressed the multifactorial nature of glaucoma, noting that there is a lot of gray area and decision-making involved, making it uniquely suited for AI and telemedicine. When you can coalesce all that information, examine it, and interpret it with an algorithm, it’s helpful in guiding the diagnosis, she said. Dr. Habash also highlighted the ability to integrate datasets that synthesize demographics, age, race, medical history, family history, etc. Those are all important factors in formulating your response to the patient sitting in front of you or on the other side of the computer screen, she said. An AI algorithm would help give direction on target pressure and figuring out risk of progression with all of these factors taken into account. She described the potential of AI as like “having a spotter in the gym,” helping integrate all the info in real time. Eventually we will be able to give our patients a real-time risk score, she said.
Dr. Habash added that with the use of AI, a patient could be tested in the optometrist’s office for high risk or high progression, which would help determine when the patient needs to be expedited to the specialist’s office. This can help prioritize high-risk patients to get them in faster for quicker intervention. A telehealth consult with the glaucoma specialist would be a natural extension in these cases.
The realm of remote monitoring combines both AI and telehealth, Dr. Habash said. There is home IOP monitoring, home OCT (coming on the market soon), visual fields for home monitoring, and fundus photography.
Dr. Habash said that there could be exponential growth when an algorithm is developed for diagnosis based on paired data from visual fields, fundus, and OCT.
Retina may be the first step, she said, because there are fewer factors involved, and it’s more of a linear relationship between an OCT and what you do for patients than in glaucoma. But that’s exactly what makes AI and telemedicine so valuable when it comes to glaucoma, Dr. Habash said.
Dr. Habash pointed to recent work she’s been doing with Microsoft looking at fundus photography of optic nerve images and OCT images to start training AI models. This type of algorithm can become more robust and reliable if we can use different datasets from around the world, she said, including different geographic backgrounds, race, ethnicities, socioeconomic situations, etc.
Dr. Habash added that there may be potential challenges with AI. “We’re just going to get better and better with time and larger amounts of data,” she said, but she added that machine learning is “like teaching a child.” You need to keep reinforcing these lessons so the machine learns better, and that will come with time and repetition, Dr. Habash said.
Dr. Trubnik also noted potential limitations, particularly being careful that you don’t have confirmation bias with such a large sample size of data. She stressed the difference between association and causation. Just because you find the association doesn’t mean it leads to something and you have to act upon it, Dr. Trubnik said.
When talking about deep learning, Dr. Trubnik said there is a “black box” for how AI comes up with a diagnosis; there may be questions from patients on how the diagnosis was determined. “You develop a relationship with the patient and explain things to them, but if the diagnosis is coming from a machine, they may question how you arrived at the diagnosis,” she said.
Dr. Trubnik added that for big data to be useful, a lot of images are needed, and sometimes there aren’t enough to yield accurate information. Also, when sharing information, she said, sometimes it’s hard to know if the data is all equal.
Additionally, she said that rare diseases may be difficult to diagnose with AI because there is not enough data to recognize them. Specifically relating to glaucoma, Dr. Trubnik said that arriving at the same definition of glaucoma may be a challenge.
“The other thing is if the AI is screening fundoscopy or OCTs, usually it’s utilizing a ‘binary classification,’ so it will say ‘yes glaucoma’ or ‘no glaucoma,’ but when I examine my patient, I can also pick up cataract, epiretinal membrane, etc.,” Dr. Trubnik said. “So far, they haven’t been able to do that accurately with AI because when they incorporate multiple variables, it’s not as accurate at predicting.”
Going back to the gym analogy, Dr. Habash said, “Technology helps us lift the weight, but we’re still the ones doing the work. AI and telemedicine won’t replace us, but they will augment us.”

At a glance

• Physicians say AI will allow them to coalesce a lot of information, examine it, and interpret it with an algorithm to help in guiding a diagnosis.
• Telemedicine also factors into AI and glaucoma, with home IOP monitoring, home OCT, visual fields for home monitoring, and fundus photography potentially playing a role.
• Possible limitations of AI in glaucoma are the complex nature of the disease and the amount of images needed in order to yield accurate information.

About the doctors

Ranya Habash, MD

Bascom Palmer Eye Institute
Miami, Florida

Valerie Trubnik, MD
Ophthalmic Consultants
of Long Island
Lynbrook, New York

Relevant disclosures

: Microsoft
Trubnik: None



Opportunities for artificial intelligence in glaucoma Opportunities for artificial intelligence in glaucoma
Ophthalmology News - EyeWorld Magazine
283 110
220 471
True, 8