Blab AI
AI Enhances Diagnostic Workflows in Ophthalmology, but Surgeons Remain Central, Experts Say
In a recent video interview, three practicing ophthalmologists explained how artificial intelligence (AI) is already embedded in routine eye care but is not yet a substitute for surgical expertise. Dr. Robert Ang, MD, emphasized that AI can process large volumes of data quickly, yet it does not perform treatments and is unlikely to replace surgeons in the near future. Dr. Radhika Rampat, a consultant in London, described specific AI‑driven tools she uses for intraocular lens (IOL) selection and keratoconus screening. Dr. John W. Kitchens, MD, outlined how early detection of geographic atrophy (GA) has evolved, with optometrists increasingly identifying subtle disease on optical coherence tomography (OCT) and autofluorescence.
AI’s current role in ophthalmology is largely diagnostic and decision‑support. According to the interview, Dr. Ang noted that AI can “process volume and speed up analysis” but is not capable of performing surgical procedures. He added that he does not expect AI to replace the surgeon anytime soon. This view aligns with the broader consensus that AI tools are designed to augment clinical judgment rather than replace it.
Dr. Rampat’s practice illustrates how AI is already part of everyday workflows. For IOL power calculation, she relies on the European Society of Cataract & Refractive Surgeons (ESCRS) online calculator, which incorporates multiple AI‑driven formulas to predict postoperative refractive outcomes. She also uses the CSO MS‑39 corneal topographer, whose built‑in AI flags potential keratoconus. The MS‑39 employs a support‑vector‑machine algorithm to classify corneal topographies as normal, suspect keratoconus, or keratoconus, achieving high sensitivity and specificity in clinical studies. While the device can alert clinicians to abnormal findings, Dr. Rampat stresses that clinical judgment remains essential.
Looking ahead, Dr. Rampat highlighted anterior‑segment applications as an emerging area. She has authored a paper on this topic, noting that most existing AI studies focus on posterior‑segment diseases such as age‑related macular degeneration. Her work suggests that AI could play a larger role in diagnosing conditions affecting the cornea, lens, and anterior chamber.
Dr. Kitchens discussed geographic atrophy, an advanced form of dry age‑related macular degeneration. He explained that his approach has shifted from treating only foveal‑involving disease to identifying early GA in patients who still have good vision and healthy retinal pigment epithelium. He credits optometrists for increasingly recognizing subtle GA on OCT and autofluorescence, which allows patients to receive treatment earlier. The video shows Dr. Kitchens wearing a fishing hat, underscoring that his clinical insights are grounded in routine practice.
The adoption of AI in eye care reflects broader trends in healthcare. AI systems can analyze imaging data faster than humans, potentially reducing diagnostic delays. However, regulatory bodies such as the U.S. Food and Drug Administration and the European Medicines Agency continue to require rigorous validation of AI algorithms before they can be marketed as medical devices. Privacy concerns also arise when large datasets of patient images are used to train models.
In summary, AI is already influencing ophthalmic practice through tools that assist with IOL selection, keratoconus screening, and early GA detection. While these systems improve efficiency and may enhance diagnostic accuracy, surgeons and clinicians remain central to patient care. Future developments will likely focus on expanding AI applications to other eye conditions, improving algorithm transparency, and ensuring compliance with regulatory standards.
The current landscape shows that AI is a valuable adjunct rather than a replacement for human expertise. Ongoing research, clinical validation, and regulatory oversight will shape how these technologies evolve in the coming years.
AI’s current role in ophthalmology is largely diagnostic and decision‑support. According to the interview, Dr. Ang noted that AI can “process volume and speed up analysis” but is not capable of performing surgical procedures. He added that he does not expect AI to replace the surgeon anytime soon. This view aligns with the broader consensus that AI tools are designed to augment clinical judgment rather than replace it.
Dr. Rampat’s practice illustrates how AI is already part of everyday workflows. For IOL power calculation, she relies on the European Society of Cataract & Refractive Surgeons (ESCRS) online calculator, which incorporates multiple AI‑driven formulas to predict postoperative refractive outcomes. She also uses the CSO MS‑39 corneal topographer, whose built‑in AI flags potential keratoconus. The MS‑39 employs a support‑vector‑machine algorithm to classify corneal topographies as normal, suspect keratoconus, or keratoconus, achieving high sensitivity and specificity in clinical studies. While the device can alert clinicians to abnormal findings, Dr. Rampat stresses that clinical judgment remains essential.
Looking ahead, Dr. Rampat highlighted anterior‑segment applications as an emerging area. She has authored a paper on this topic, noting that most existing AI studies focus on posterior‑segment diseases such as age‑related macular degeneration. Her work suggests that AI could play a larger role in diagnosing conditions affecting the cornea, lens, and anterior chamber.
Dr. Kitchens discussed geographic atrophy, an advanced form of dry age‑related macular degeneration. He explained that his approach has shifted from treating only foveal‑involving disease to identifying early GA in patients who still have good vision and healthy retinal pigment epithelium. He credits optometrists for increasingly recognizing subtle GA on OCT and autofluorescence, which allows patients to receive treatment earlier. The video shows Dr. Kitchens wearing a fishing hat, underscoring that his clinical insights are grounded in routine practice.
The adoption of AI in eye care reflects broader trends in healthcare. AI systems can analyze imaging data faster than humans, potentially reducing diagnostic delays. However, regulatory bodies such as the U.S. Food and Drug Administration and the European Medicines Agency continue to require rigorous validation of AI algorithms before they can be marketed as medical devices. Privacy concerns also arise when large datasets of patient images are used to train models.
In summary, AI is already influencing ophthalmic practice through tools that assist with IOL selection, keratoconus screening, and early GA detection. While these systems improve efficiency and may enhance diagnostic accuracy, surgeons and clinicians remain central to patient care. Future developments will likely focus on expanding AI applications to other eye conditions, improving algorithm transparency, and ensuring compliance with regulatory standards.
The current landscape shows that AI is a valuable adjunct rather than a replacement for human expertise. Ongoing research, clinical validation, and regulatory oversight will shape how these technologies evolve in the coming years.
