Read more

May 05, 2023
1 min read
Save

AI calculator helps select IOLs for short eyes

You've successfully added to your alerts. You will receive an email when new content is published.

Click Here to Manage Email Alerts

We were unable to process your request. Please try again later. If you continue to have this issue please contact customerservice@slackinc.com.

Key takeaways:

  • The Zeiss AI IOL calculator had a smaller mean absolute error than other IOL formulas.
  • The model is based on 500,000 simulated eyes as well as real-world data.

SAN DIEGO — An artificial intelligence-based IOL calculator may help optimize models for different kinds of eye parameters, including short eyes, according to a study.

When conducting an IOL calculation, optimizing the IOL constant is important if the cohort in the study is representative of the overall population, Douglas D. Koch, MD, said at DOS Digital Day at the American Society of Cataract and Refractive Surgery meeting.

Douglas D. Kotch, MD

“However, optimizing the constant can be misleading or unhelpful if the cohort is not representative of the population at large, for example, short eyes,” he said. “The constant optimized for that cohort may not be helpful for other surgeons and colleagues.”

Koch presented data on the Zeiss artificial intelligence IOL calculator, the first AI-based and IOL-specific formula.

In addition to biometry measurement and target refraction, Koch said the Zeiss AI IOL calculator uses optimized information for each unique IOL model rather than using an IOL constant.

He said the model starts with paraxial ray tracing and a database of 500,000 simulated model eyes that it uses to predict IOL power. From there, preoperative biometry and postoperative anterior chamber depth are used to predict the postoperative IOL position. It also includes curated data from Zeiss’ Veracity surgery planner data collection.

Koch said in a study comprising 155 single eyes and 123 paired eyes, the Zeiss AI IOL calculator had a smaller mean absolute error than other calculators, such as Kane, Barrett and Pearl-DGS (P < .05). It also had a smaller root mean square absolute error.

“The future for this is that you can build up this database with eyes from a wide range of parameters — anterior chamber depth, lens thickness, corneal diameter — and optimize for each model,” Koch said.