ESCRS - FP27.02 - Influence Of Lens Diameter On Postoperative Intraocular Lens Position

Influence Of Lens Diameter On Postoperative Intraocular Lens Position

Published 2024 - 42nd Congress of the ESCRS

Reference: FP27.02 | Type: Free paper | DOI: 10.82333/0cpt-xs13

Authors: Sophia Anna Reifeltshammer* 1 , Leon Pomberger 1 , Lisa Tasch 1 , Klemens Waser 1 , Matthias Bolz 1 , Nino Hirnschall 1

1Ophthalmology and Optometry,Johannes Kepler University,Linz,Austria;Ophthalmology and Optometry,Kepler University Hospital,Linz,Austria

Purpose

To investigate the effect of preoperative lens diameter (LD) as well as preoperative anterior chamber depth (ACD) on postoperative intraocular lens (IOL) position and refraction using an artificial intelligence approach.

Setting

Department of Ophthalmology and Optometry, Kepler University Hospital, Linz, Austria

Department of Ophthalmology and Optometry, Johannes Kepler University, Linz

Methods

This monocentric study included patients that underwent standard cataract surgery implanting a monofocal hydrophobic open loop haptic IOL. Preoperative measurements included optical biometry (IOLMaster700, Carl Zeiss Meditec AG, Germany) and anterior segment optical coherence tomography (AS-OCT, Casia2, Tomey, Japan). Study examinations were performed 6 weeks after surgery and included AS-OCT, optical biometry, autorefraction as well as subjective refraction. Random forest plot and a partial least squares regression (PLSR) model were used to predict the influence of ACD and LD on the postoperative IOL position.

Results

A total of 400 eyes were included in this study. Preliminary data show that preoperative ACD (mean 3.07 ± 0.40 mm), LD (mean 10.18 ± 0.36 mm), and axial eye length (mean 23.53 ± 1.41 mm) were the main predictors of postoperative anterior chamber depth. Random forest plot was then trained using these three parameters (out of bag: 0.040). The ACD predicted by the machine learning algorithm as well as the measured LD were significantly correlated with the postoperative ACD.

Conclusions

ACD and LD can be effectively used to train a machine learning model to predict postoperative IOL position. A more thorough analysis will be presented at the meeting.