AN AUTOMATED CATARACT DETECTION SYSTEM USING DEEP LEARNING FOR FUNDUS IMAGES

UDC 681.51
DOI: 10.36871/2618-9976.2023.01.009

Authors

Eugene Yu. Shchetinin,
Doctor phys.-math sciences, professor, Financial University under the Government of Russian Federation, Moscow, Russia

Abstract

Cataract is one of the most common eye diseases that causes serious visual impairment. Accurate and timely detection of cataracts is the most effective way to prevent the onset of blindness. The article implements an automatic cataract detection system on a publicly available dataset of fundus images ODR using deep learning methods. For this purpose, a model of classification of fundus images based on the deep neural network VGG19 has been developed.
On the publicly available set of fundus images ODIR, the accuracy of cataract detection (accuracy_class) using this model was 97,23%, precision= 99,11%, sensitivity= 97,12%.

Keywords

Cataract, Deep learning, Convolutional neural networks