Researchers have developed deep-learning (AI) models to support ophthalmologists in diagnosing corneal infections, such as keratitis (IK). This inflammation of the cornea affects millions of people worldwide, especially in developing countries where access to specialized eye care is limited. Keratitis is therefore an important cause of corneal blindness on a global level.
The study, which analyzed 35 studies on the use of AI models to diagnose infectious keratitis, reports good results. Those are published in eClinicalMedicine, part of The Lancet.
AI performs slightly better
The research concludes that the AI models actually perform as well, or actually fractionally better, than ophthalmologists. For the study, a total of more than 136,000 images of the cornea were analyzed by the various AI models. The AI models matched the diagnostic accuracy of ophthalmologists with a sensitivity of 89.2 percent and a specificity of 93.2 percent. The ophthalmologists achieved 82.2 percent sensitivity and 89.6 percent specificity, respectively, during diagnosis.
“Our research shows that AI has the potential to provide rapid, reliable diagnoses, which could revolutionize the way we treat corneal infections worldwide. This is particularly promising for regions where access to specialist eye care is limited, and could help reduce the burden of avoidable blindness worldwide,” said Dr. Ting, lead author of the study, Birmingham Health Partners (BHP) Fellow and consultant in ophthalmology at the University of Birmingham.
Distinguish underlying causes
The AI models also proved effective in distinguishing between healthy eyes, infected corneas and the various underlying causes of IK, such as bacterial or fungal infections.
While these results demonstrate the potential of deep learning and AI models in eye care, researchers also emphasized that, in addition to external validation of the findings, more diverse data are also needed to increase the reliability of these models for clinical use. As AI technology continues to grow and play a central role in medicine, it may soon become an important tool in preventing corneal blindness worldwide.
Glaucoma in Pictures
Research has also been conducted in the Netherlands for several years into the use of AI models for better (and faster) diagnosis of eye disorders. A good example is the AI project ‘Glaucoma in Image’. More than 100,000 fundus photos were made available for this purpose in 2020. These photos were used to ‘train’ the computers to recognize glaucoma using AI. For this purpose, all fundus photographs were first assessed by two specialists. That data, together with the photos, was entered into the computer system.
It was then checked whether the computer made the same diagnosis as the specialists. Due to the enormous amount of photos and reviews, the computer, with the help of AI (artificial intelligence), will learn to determine from a fundus photo whether the eye in question has glaucoma.
AI in Eye Care: A Cheeky Look at the Future of Ophthalmology
Well, folks, it looks like the future is here, and it’s got a rather impressive pair of specs on! Researchers have developed deep-learning AI models that are stepping in to help ophthalmologists diagnose corneal infections, such as keratitis (we won’t be asking for a spelling test, don’t worry). This inflammation is no small matter; it affects millions worldwide, particularly in developing countries where a visit to the eye doctor can be as rare as finding a needle in a haystack. And given what we know about keratitis, it’s a significant culprit behind worldwide corneal blindness. Yikes!
Published in eClinicalMedicine (not exactly the healthy eating guide we all need, right?) the study diligently sifted through 35 other studies focusing on AI’s prowess in diagnosing infectious keratitis. And guess what? The results were more promising than finding a twenty-pound note in last winter’s coat pocket!
AI is Not Just Good—It’s Better!
Surprisingly, the AI models outperformed human eyeballs when it came to diagnostic accuracy. Imagine that! With a sensitivity of 89.2 percent and specificity of 93.2 percent, these models certainly have a knack for spotting trouble faster than the average doctor can say, “Have you tried taking some over-the-counter drops?” By contrast, ophthalmologists achieved 82.2 percent sensitivity and 89.6 percent specificity. So, what we’re really saying is, if you thought AI couldn’t see the wood for the trees, you might want to reconsider!
Dr. Ting—with a name that could easily belong to either a superhero or a wildly confused eye doctor—expresses that AI has the potential to provide rapid, reliable diagnoses. He claims this could revolutionize our approach to treating corneal infections. Well, that’s a bold statement, Doctor! If AI is doing it faster and better, does that mean we might one day have to swap our eye charts for a computer screen? Who would’ve thought it possible?
Let’s Get to the Root of the Issue!
But hold your horses! The AI didn’t just stop at diagnosing; it also distinguished between healthy eyes, infected corneas, and the underlying causes of keratitis, be it bacterial or fungal. It’s almost like AI models are the Sherlock Holmes of the eye world! But of course, before we throw a parade, the researchers did emphasize that we need more diverse data to validate these findings further. Apparently, the computer’s not quite ready to graduate just yet!
As this AI wave rolls on, it seems we may find ourselves welcoming these digital eye experts into clinics worldwide. After all, preventing corneal blindness on a global scale sounds like a cause we can all get behind. Whether you’re in a bustling city or a remote village, nobody should have to battle the blurry world without help!
Glaucoma? No Problem!
And to add another feather in the cap of AI in eye care, research from the Netherlands is taking things a step further with the aptly named ‘Glaucoma in Image’ project. They’re using AI to spot glaucoma faster than you can say “don’t forget your eye test.” By training a computer on over 100,000 fundus photos (or what we lovingly call ‘pictures of the back of your eyeball’), specialists are allowing AI to learn and identify glaucoma with precision down to the pixel! Now that’s what I call seeing things clearly.
So, there you have it! With AI becoming an essential toolkit for ophthalmologists, we may well be on the brink of a revolution in eye care. But let’s just hope for everyone’s sake that the robots don’t start plotting world domination. Who wants a cyborg looking deep into your soul while they check if you’re well-hydrated!?
Here’s hoping our future includes fewer trips to eye doctors and more streamlined, accurate diagnoses. After all, folks, there’s a whole world out there—and let’s face it, a lot of it looks a bit blurry right now.
