Dr. Francis Mah Discusses AI for Diagnosing Anterior Segment Conditions, AS-OCT for Surgical Planning, and Continued Unmet Needs

Francis Mah, MD, of the Scripps Clinic in La Jolla, California, discusses the use of artificial intelligence in eye care and continued unmet needs for anterior segment conditions.

Question:

Can you talk about the use of AI in detecting and managing anterior segment diseases? How will this impact the field?

Francis S. Mah, MD:

Artificial intelligence, I think it’s been this wave that’s been happening and I think now it’s getting a lot more notoriety and I think it’s, I don’t want to say ready for prime time, but it’s almost ready for prime time. I think basically, there are a couple of factors which really are important in terms of artificial intelligence.

Number 1 is the amount of data that we have or that big data. The more data that we can generate, the better our outcomes would be in general because then it would account for even the small variations that can exist. I think the second thing is who is kind of describing the algorithm that the artificial intelligence needs to use for the big data? It’s kind of like the clinicians or the programmers working together to follow this algorithm or follow the questions that artificial intelligence needs to be asked and then artificial intelligence needs to answer. It really depends, and I’m sure everybody’s heard of examples of how changing the question can change the answers that are fit out by artificial intelligence.

I think it’s not that in the future eye care specialists are going to be out of a job. I think they are going to be important. Some of them are going to be very important in determining the correct questions and the algorithms that are used by artificial intelligence. I think the easiest things in terms of anterior segment and the biggest impact is going to be glaucoma. Obviously, with the huge number of glaucoma patients that are out there, huge amounts of data are being generated and they’ve been generated over the years with visual fields as well as OCTs. This generation of data can more easily be used for the development of these protocols for artificial intelligence. You can put them all together with pressure and age and everything that we kind of ask today. We have a lot of that data and so I think that’s where I think the soonest impact is really going to be made.

The other areas is refractive surgery. Obviously, again, there’s a lot of data that’s out there as far as pre-op, post-op, even complications. When to avoid surgery to avoid ectasia, for example. When is it going to be dangerous? When is it safe? Cataract surgery, another huge area where you’ve got a lot of biometry, a lot of surgeons. We know the pre and post measurements and I think it’s already being used. There are a couple of formulas that we’re using to intraocular lenses around cataract surgery that are basically using AI. The Hill-RBF, for example, is a great example. That is, it only improves as you use it more and more data is put in there. I think there’s a lot of areas that are going to be impacted very, very soon. That are being impacted that we don’t even know and every day, every week they’re being refined as far as AI. Again, it’s not a brand new thing. It’s just being refined and it’s really come to the spotlight nowadays.

Question:

Can you talk about the use of anterior segment optical coherence tomography for surgical planning as well as detecting ocular changes post-surgery?

Francis S. Mah, MD:

The anterior segment OCT, think it helps in very obvious ways in two primary conditions. Number 1 is we kind of talked about it before, the ocular surface neoplasias. It helps differentiate them. It also helps define the areas and as far as the margins for example. I think that really helps the planning stages and if you can plan for the margins and you can make the diagnosis sooner, that’s obviously going to impact the outcomes much better.

As far as the other aspect, that’s the internal aspect. It’s for example, DSEK/DMEK. Right now, very easily it’s being used if you have anterior segment OCT in the operating room, for example, for DMEK surgery to tell whether the scroll is in the proper orientation or if it’s upside down. It’s pretty easy to take a picture and then for the surgeon to proceed in a manner that will benefit the patients more expeditiously than just playing around with the tissue or thinking that you’ve got the correct orientation and then finding out later, whether it’s intra-op or post-op, that the orientation was not accurate.

Then I think in terms of visualization, sometimes it’s a little difficult when you’ve got, for example, a cloudy cornea or some scars. You can kind of dim the lights and then rely on anterior segment OCT a little bit better to help with the internal structures of the eyes and pathology. It also should help with MIGS and some of the other glaucoma procedures as well in terms of identifying the planes that you want to be in and where exactly are your devices. I think in general, these are the areas and especially as the resolution continues to improve, and then the area that we’re actually able to visualize increases, I think those are the first areas that are going to really benefit from anterior segment OCT.

Question:

What are some areas of continued unmet need or research that will be important to advance the field of anterior segment diseases?

Francis S. Mah, MD:

I think a couple of areas. One, we kind of mentioned presbyopia. Presbyopia, there are some good ways to manage it. Obviously glasses, contact lenses. Obviously, monovision with the contact lenses. With cataract surgery, we’ve got multifocal IOLs, we’ve got EDOF lenses, we’ve got monovision that we could do. But we still don’t have, for example, a reversible way of managing topically let’s say with eye drops. It’s either glasses or contacts or surgery. It would be great to have a very safe, very effective treatment and maybe a permanent solution for presbyopia. I think that would be huge.

As far as refining, it’s improving, but there’s always this question of does a patient have glaucoma? Does a patient not have glaucoma? Sometimes there’s time that’s passing where we would like to jump on this sooner if it is glaucoma or we’d like to reassure the patient that they don’t have glaucoma. I think that is an unmet need in terms of earlier diagnosis.

As far as keratoconus, we’ve come a long way with an ectasia, post-refractive ectasia. We’ve come a long way with cross-linking. That’s a phenomenal technology that probably should be used more often. But in terms of actually improving vision with patients that have ectasia and keratoconus, we still, I think, are lacking in that area. The cross-linking is great for stopping. It’s not really improved and indication wise for improvement in vision. We still need something, whether that’s to be used in conjunction with eczema lasers for keratorefractive procedure for example, that I think is still being refined.

I think somebody’s kind of come up with cataract drops or something like that, that are going to treat cataracts before they form. I think in terms of the implant lenses, the challenge and the advances in IOLs that are like a 20-year-old eye that can see distance up close and everything in between with no dysphotopsias or issues with the quality of your vision, I think that is improving, but obviously we’re not there yet.

As far as infections, we still don’t have an antiviral for adenovirus, so that would be nice. There’ve been a couple that are in clinical trials, but we still don’t have anything there that can really address that from an asset or a medication standpoint. It would be nice to have something that could do that. Finally, I think scar modification or even pain management could be a little bit better. I think dry eye management and issues with ocular surface disease, I think that could also be improved. Although again, there have been huge advances in the past even five years. There are some areas of unmet need still that are out there and I’m sure that there’s other smart people out there that have figured out some other areas of unmet need that I’m not mentioning now.