An Integrated Diabetic Retinopathy Index for the Diagnosis of Retinopathy Using Digital Fundus Image Features

摘要

Retinal fundus images may provide useful insights into the visual system and increasingly, provide additional indicators of systemic health. Many important eye and systemic diseases and conditions may manifest themselves as morphological changes that may be observed and recorded in the retina. Some of the most prevalent causes of blindness in the world include age-related macular degeneration, Diabetic Retinopathy (DR), and glaucoma. As many of these conditions are often asymptomatic till the more advanced stages, systematic screening of targetted individuals may help to identify the cases for early and appropriate intervention. This paper proposes a system for an automated identification of normal and abnormal DR classes using digital fundus images. Local Binary Pattern (LBP), Entropies and Invariant moments were used to extract the salient features. In this work, we have proposed a novel integrated index called Diabetic Retinopathy Index (DRI) which is made up of different features, to diagnose the unknown class using a single number. We hope that this DRI can aid clinicians towards faster screening of people with diabetes.