CLASSIFICATION OF MULTIPLE EYE DISEASES USING RETINAL FUNDUS IMAGES
Abstract
Retinal fundus images have an important role in detection, tracking and treatment of different eye abnormalities. Early diagnosis and treatment of such eye abnormalities will help in preventing vision loss. Some of the commonly seen eye abnormalities include diabetic retinopathy (DR), optic disc cupping (ODC), media haze (MH), cataract and age-related macular degeneration (ARMD). This paper proposes an automated CNN based deep learning model for detection and classification of eye abnormalities. RFMiD is a publicly available dataset containing 3200 fundus images with 45 different eye abnormalities. A multi-layer deep neural network has been developed to train and test images of different eye abnormalities. This paper focuses on classification of the retinal eye diseases, DR, ODC, ARMD. This model outperforms other current models and demonstrates its effectiveness in identifying and categorizing retinal eye disorders.