Computer-Based Identification of Diabetic Maculopathy Stages Using Fundus Images

Diabetes mellitus is a major cause of visual impairment and blindness. Almost all patients with type 1 diabetes and over 60% of patients with type 2 diabetes will develop some degree of retinopathy after having diabetes for 20 years. Prolonged presence of diabetic retinopathy will result in maculopathy, a condition in which the macula of the eye gets damaged. Depending on the extent of maculopathic damage, vision will become progressively impaired. This chapter presents an intelligent system that is capable of differentiating between fundus eye images belonging to normal eyes and those belonging to eyes affected by either of the two types of maculopathy – nonclinically significant macular edema (nonCSME) and clinically significant macular edema (CSME). In this proposed technique, characteristic features were extracted from the raw fundus images using morphological image processing techniques and fed to a feed-forward artificial neural network (ANN). Ninety subjects belonging to normal, nonCSME, and CSME categories were used for evaluating the proposed technique. Results showed that the p-values (obtained using the analysis of variance (ANOVA) test) of the selected features were less than 0.001, which indicates that the features are clinically significant and very discriminative. The ANN classifier resulted in a good accuracy of 96.67%, and sensitivity and specificity of 96.67 and 100%, respectively.