Predicting Post-Operative Intraocular Lens (Iol) Tilt Using Preoperative Optical Biometry Data And A Machine Learning Algorithm

Published 2022 - 40th Congress of the ESCRS

Reference: FPT08.05 | Type: Free paper | DOI: 10.82333/hw38-nz42

Authors: Klemens Waser* ¹ , Nino Hirnschall ¹ , Peter Laubichler ¹ , Haidar Khalil ¹ , Paul Jirak ¹ , Leon Pomberger ¹ , Matthias Bolz ¹

¹Ophthalmology,Kepleruniversitätsklinik Linz,Linz,Austria

Purpose

Purpose: Intraocular lens (IOL) tilt reduces visual performance, if it exceeds the amount of physiological tilt. Furthermore, tilt mimics astigmatism and influences the performance of toric IOLs as well as aspheric IOLs. Aim of this study was to use pre-operative IOL measurements to predict the post-operative amount and orientation of tilt.

Setting

In total, 200 eyes scheduled for cataract surgery are included in this prospective study. The patients are recruited at the Kepleruniversitätsklinikum Linz.

Methods

Preoperatively, anterior segment imaging will be performed using the IOL Master 700 (Carl Zeiss Meditec AG, Germany), Casia 2 (Tomey, Japan), MS-39 (CSO, Italy) and Pentacam HR (Oculus, Germany). These measurements are repeated 8 weeks after surgery. At this visit, autorefraction and subjective refraction will be performed additionally. Data will be analysed a machine learning algorithm to predict post-operative tilt.

Results

Preliminary pre-operative data showed an IOL tilt of the crystalline lens of 4.9° (SD: 1.4°, Median 4.9°). The data was normally distributed. In 90% of the cases, the orientation of the IOL tilt was inferotemporal, 10% (n=2) horizontal. The prediction algorithm showed good predictability of the orientation of tilt, but a relevant imperfection regarding the amount of tilt.

Conclusions

The post-operative orientation of tilt is predictable, whereas the amount of tilt is only partly predictable. These findings should help to improve IOL power calculation concepts, especially for toric IOLs.