Detection of Abnormal Features in Digital Fundus Image Using Morphological Approach for Classification of Diabetic Retinopathy

Over the past few years diabetic retinopathy has proved to be a leading cause of blindness in adult population. WHO studies in 2002 showed that the diabetic retinopathy accounts for nearly 5% of the 37 million visually challenged people worldwide. The number of diabetic retinopathy patients around the globe is expected to increase from 127 million in 2010 to 191 million in 2030. An innovative and effective mass screening system needs to be developed for meeting the ever increasing number of patients with the limited number of ophthalmologists. The digital fundus image of retina is being effectively used for the diagnosis of diabetic retinopathy. Microaneurysms, haemorrhages and exudates are the abnormal features commonly observed in the retinal image of a person affected by diabetic retinopathy. In this paper a method using mathematical morphological operations and edge detection to detect these lesions is discussed. The detected features are used to classify the different stages of diabetic retinopathy. The method discussed is fast and robust (applicable on low quality images) and hence suitable for mass screening of patients.