Intraocular Lens Calculation Using Artificial Intelligence

Overview

The aim of this study is to investigate the lens diameter (LD) as a useful parameter in intraocular lens (IOL) power calculation by using several different non-invasive imaging techniques. The aim is to establish an accurate model for lens diameter estimation and subsequently evaluate the influence of the LD on the effective postoperative lens position. The comparability of the different devices in terms of variability between the biometric measurements will also be assessed. By then combining these two approaches with artificial intelligence, the aim is to develop a new approach to effectively incorporate the LD in IOL power calculation and improve patient's outcome in the long run.

During uncomplicated cataract operation, the clouded natural lens is being removed and replaced with an artificial intraocular lens (IOL). To achieve optimal postoperative outcomes for patients, the power of the implanted artificial lens is calculated prior to surgery. Significant progress has been made in calculation concepts over the past 20 years, including the introduction of regression, vergence and ray tracing. More recently, approaches incorporating artificial intelligence have emerged. All these formulae are based on the biometric data of the eye. This includes parameters such as axial eye length, corneal curvature, central corneal thickness, anterior chamber depth and the refractive indices of the eye's optical segments. By including all these variables, modern formulas aim to deliver the best possible postoperative outcomes. One variable that has not been included in the calculation thus far is the diameter of the natural lens. Large parts of the lens are covered by the iris. Even with medically dilated pupils the peripheral parts cannot be visualized, and subsequently not adequately reproduced using established imaging methods. This has made implementation in IOL power calculation difficult. In everyday clinical practice, however, anterior segment OCT imaging devices are equipped with features that allow for an estimation of lens diameter. This is achieved by extrapolating the anterior and posterior curvature of the natural lens, which unfortunately makes this approach prone to error. Other imaging techniques, such as magnetic resonance imaging, are impractical in routine clinical practice due to time and cost considerations. However, they could be highly beneficial for future predictive approaches of the lens diameter The aim of this study is to develop a model for incorporating the lens diameter into IOL calculation. This will be achieved by using different imaging technologies to determine the actual lens diameter. The diameter will then be predicted using available biometric variables.