Growing Options Optimise Outcomes in Cataract Surgery

The array of IOL power formulas has expanded through an influx of new options based on artificial intelligence (AI) or incorporating some AI elements. Studies evalu­ating the performance of these AI-driven formulas show they hold potential promise for providing more accurate refractive outcomes and might be especially useful in more challenging situations where refractive outcomes tend to be more unpre­dictable, i.e., when applied in cases involving short, long, or post-refractive surgery eyes. 

However, more studies evaluating AI-driven formulas are needed because they are still considered “young technology”, said Thomas Kohnen MD, PhD.

Further advances might depend on the remaining need to unlock the solution for attaining improved prediction of effec­tive lens position (ELP), according to Oliver Findl MD.

“We know the refractive outcome is probably the most critical of all outcome parameters for achieving patient sat­isfaction after cataract surgery,” he said. “Available modern formulas perform very well in normal eyes, but consistently achieving the target refraction is more challenging in short and long eyes, where the main issue is accurate prediction of IOL position after surgery.”

Until there are advances such that IOL power can be con­sistently selected to achieve the targeted refractive outcome, perhaps the formulas should be renamed IOL power estima­tion formulas, Dr Kohnen observed. 

“Calculation may be the wrong descriptive term because it implies that we always get the right target, but we don’t,” he explained. “Rather, what we have with current formulas is mostly an approximation or estimation.”

Classifying the AI options

The proliferation of AI-driven IOL formulas is highlighted in a forthcoming article in the Journal of Cataract & Refractive Surgery authored by Drs Giacomo Savini, Ken Hoffer, and Thomas Kohnen, outlining a new classification system for IOL power formulas. Their scheme groups them into three broad categories: vergence, AI, and raytracing. The AI category includes the Hill RBF, Karmona, and Nallasamy formulas. However, subdivision of the vergence and raytracing for­mulas based on additional features shows the inclusion of formulas with AI elements.

The only “AI” formula within the raytracing category is the Zeiss AI IOL calculator, but the vergence category includes five—namely the thick lens PEARL-DGS and the 3C Calculator, Hoffer QST, Kane, and Ladas AI, which are all thin lens formulas.

Published studies reviewed by Drs Findl and Kohnen highlight good predictive accuracy of the AI-driven formulas. The differences in performance were sometimes small and perhaps not clinically relevant. However, when considering outcomes across the full spectrum of short, normal, and long eyes, the AI-based approaches appear to provide the best results overall, Dr Kohnen said.

Discussing the need for better prediction of ELP, Dr Findl also cited a study that found strategies using deep learning algorithms outperformed a traditional multilinear regression technique.

Easing the planning burden

As the number of IOL power formulas continues to grow—and with many options accessible only online and not on instru­ments surgeons use routinely for preoperative biometry— checking and comparing the results generated by different formulas can become tedious and time-consuming. The online ESCRS IOL calculator represents a useful tool addressing this challenge because it allows surgeons to efficiently access results from multiple modern IOL formulas after inputting data just one time into a single entry form.

“The ESCRS calculator is a convenient tool that includes the most recent and better IOL formulas and is something cataract surgeons may want to use in their daily routine, especially in the more difficult cases involving short and long eyes,” Dr Findl said.

Launched in September 2022 at the ESCRS Congress, the cal­culator originally included only standard non-toric IOL power calculation formulas. Recently, however, a toric calculator was added, and work is underway to create an option for IOL calcu­lations in eyes with a history of corneal refractive surgery.

Acknowledging the problems

Excitement about the potential for achieving improved refrac­tive results by using AI to develop IOL power formulas must be tempered by consideration of the issues accompanying its use, Dr Kohnen said.

The need for a huge amount of data underlies one chal­lenge, accompanied by concerns about security to protect the safety of digital patient data. In addition, transparency is need­ed for gaining user trust, but transparency is not yet there.

“With AI, we have a black box situation because it is not known how the results were reached,” Dr Kohnen said. “Im­provements are needed for understanding the decision-mak­ing process of the AI models.”

Drs Findl and Kohnen spoke at the ESCRS Symposium during AAO Refractive Surgery Subspecialty Day 2023 in San Francisco, US.

Oliver Findl MD, MBA is Chief of the Department of Ophthalmology, Hanusch Hospital, Vienna, Austria. ofindl@gmail.com

Thomas Kohnen MD, PhD, FEBO is Professor and Chair of the Department of Ophthalmology, Goethe University, Frankfurt, Germany. kohnen@em.uni-frankfurt.de

, ncluding surgical videos, diagrams, animations, quizzes, and forums.