Brave New World of AI

Artificial intelligence will play an increasing role in clinical and surgical practice in the future of anterior segment surgery, according to Andrzej Grzybowski MD, PhD.

“There has been a huge increase in the literature related to AI and ophthalmology in the last ten years,” he said. In the present debate on the issue, AI in healthcare is expected to go where no physician has before. It will be used to recognise, diagnose, and even treat conditions that presently are difficult to deal with.

AI applications in cataract surgery now include cataract screening and diagnosis, the optimal calculation of IOL power, the classification of surgical phases from videos, the prediction of surgical timing to optimise OR workflow, and PCO risk prediction.

As Professor Grzybowski explained, these applications can be divided into screening primary care, intraoperative care, and postoperative care and trained using different image modalities like anterior OCT scans, slit lamp images, or retinal fundus images.

He noted several important studies regarding AI in cataract surgery in recent years demonstrate how its applications can become increasingly developed to support the ophthalmologist in clinical practice. The first study he shared showed a free AI model can diagnose nuclear sclerosis and cortical cataracts with the accuracy slightly superior to ophthalmologists while providing similar accuracy in cases of posterior subcapsular cataracts. Another study used the same application to detect cataracts through retinal images.

An interesting application of AI is in the phase segmentation of cataract surgery. A 2022 study showed the possibility of an AI-based intraoperative guidance tool for phacoemulsification in cataract surgery, where the AI tracked pupil location in real time and computer vision provided a template for capsulorhexis, adjustable by the user. Another study on this matter last year demonstrated the deep learning system could give automated feedback to help surgeons evaluate their performance in cataract surgery. A similar Chinese study showed how a deep learning system—which gives the correct chronological order and timely alerts about incorrect workflows—has helped three residents improve their performances in cataract surgery.

Finally, other studied applications of AI included machine learning models used to predict posterior capsular rupture, actually decreasing its rate, and an AI model used in telemedicine follow-up assessment of cataract surgery patients, which demonstrated outcome sensitivity of 94% and showed moderate to strong agreement with clinicians in all parameters, asking about five key symptoms, including redness, pain, vision issues, new floaters, and flashing lights.

So, AI will be increasingly present in the future of ophthalmology, with more studies designed and new models developed. “We can expect the performance of such models to improve,” Prof Grzybowski concluded.

Prof Grzybowski presented these findings at the 2024 ESCRS Congress in Barcelona.

Andrzej Grzybowski MD, PhD is Chair of Ophthalmology at the University of Warmia and Mazury, Olsztyn, Poland, and Head of Institute for Research in Ophthalmology, Foundation for Ophthalmology Development, Poznań, Poland. He is the founder of AI in the Ophthalmology Society (https://iaisoc.com/) which now has more than 1,400 members. Membership is free. He can be reached at ae.grzybowski@gmail.com.