Posters

Prediction of postoperative effective lens position in cataract surgery using a genetic algorithm

Poster Details

First Author: A. Tamaoki JAPAN

Co Author(s):    T. Kojima   Y. Tanaka   K. Tanaka   T. Kaga   K. Ichikawa        

Abstract Details

Purpose:

To evaluate the predictive accuracy of effective lens position (ELP) by using a genetic algorithm (GA) in patients with cataract.

Setting:

Ninety-six eyes of 96 patients were included. Department of Ophthalmology, Japan Community Healthcare Organization Chukyo Hospital, Nagoya, Japan

Methods:

Axial length (AL), lens thickness (LT), anterior aqueous depth (AQD), and central corneal thickness (CCT) were measured using Swept Source Biometry (IOLMaster700,Carl Zeiss Meditec). Angle-to-angle width, angle-to-angle depth (ATA-D), and anterior and posterior corneal curvature were measured using anterior segment optical coherence tomography (CASIA, TOMEY) before and 3 months after cataract surgery. Measurements of the first 55 eyes were used to create an equation with GA minimizing postoperative ELP error. The equation was applied to the next 41 eyes to verify predictive accuracy. Stepwise multiple regression (SMR) analysis was compared with GA. AN6KA (KOWA) was implanted in all patients.

Results:

A predictive equation from GA analysis used 10 parameters including age and sec; ATA-D, AQD, LT, AL, and CCT were independent variables for SMR. Maximum error between measured ELP and predictive values was 0.398 mm for GA and 0.401 mm for SMR; average absolute error was 0.108±0.09 mm for GA, and 0.120±0.10 mm for SMR. Average absolute error for GA was significantly smaller than that for SMR (p=0.0337: Wilcoxon matched-pairs signed rank test). GA and SMR correctly predicted ELP of 56.1% and 51.2% to ±0.1 mm. In both groups, ELP of 97.6% eyes was correctly predicted to ±0.3 mm.

Conclusions:

GA analysis is a promising method for ELP prediction, and may help accurately calculate intraocular lens power.

Financial Disclosure:

NONE