Customised Selection of the Haptic Design in C-Loop Intraocular Lenses Based on Deep Learning.

Abstract

In order to increase the probability of having a successful cataract post-surgery, the customisation of the haptic design of the intraocular lens (IOL) according to the characteristics of the patient is recommended. In this study, we present two prediction models based on deep neural networks (DNNs). One is capable of predicting the biomechanical stability of any C-loop IOL, whereas the other can predict the haptic design that fits a desired biomechanical response, enabling the selection of the optimal IOL as a function of the IOL diameter compression. The data used to feed the networks has been obtained from a validated finite element model in which multitude of geometries are tested according to the ISO 11979-3 compression test, a standard for the mechanical properties of the IOLs. The biomechanical response model provides a very high accurate response (Pearson's r = 0.995), whilst the IOL haptic design model shows that several IOL designs can provide the same biomechanical response (Pearson's r = 0.992). This study might help manufacturers and ophthalmologists both analyse any IOL design and select the best IOL for each patient. In order to facilitate its application, a graphical user interface (GUI) was created to show the potential of deep learning methods in cataract surgery.