JIITA, vol. 2, no. 1, pp39-49, Mar, 2018
DOI: 10.22664/ISITA.2018.2.1.39
Automated Diabetic Retinopathy Detection using Deep Learning
Bismita Choudhury , Patrick Hang Hui Then , Valliappan Raman
Centre for Digital Futures & Faculty of Engineering, Computing and Science Swinburne University of Technology Sarawak Kuching, Malaysia
Abstract: Diabetic retinopathy (DR) is a retinal blood vascular disease caused due to diabetes mellitus. Being the leading cause of blindness, it becomes utmost important to detect DR. There is variety of reports on automatic detection of DR and most of them are dependent on the segmentation and feature extraction algorithms of various clinical signs of DR. The classification results of such methods relay on the performance of segmentation and feature extraction methods. In this paper, we have proposed a deep learning based approach to automatically detect DR. We exploited the architecture of classical Convolutional Neural Network (CNN) to learn the features of DR from the color fundus image. The convolution layers in the CNN learn the normal and abnormal features from the retina image itself. This proposed twoclass classification approach using CNN detects the DR and normal images with a high accuracy of 98.7%.
Keyword: retina; diabetes; retinopathy; deep learning; convolution
Full Paper: