Classification of Retinal Diseases from OCT scans using Convolutional Neural Networks

Abstract

Biomedical image classification for diseases is a lengthy and manual process. However recent progresses in computer vision has enabled detection and classification of medical images using machine intelligence a more feasible solution. We explore the possibility of automated detection and classification of retinal abnormalities from retinal OCT scan images of patients. We develop an algorithm to detect the region of interest from a retinal OCT scan and use a computationally inexpensive single layer convolutional neutral network structure for the classification process. Our model is trained on an open source retinal OCT dataset containing 83,484 images of various tunnel disease patients and provides a feasible classification accuracy.