Publication

Abstract : Diabetes has become one of the prominent diseases all over the world and the WHO report projects diabetes as the seventh leading cause of death and has anticipated that 366 million people will be diabetic by 2030. Prolonged diabetes causes diabetic retinopathy as well as diabetic maculopathy which are the primary reason for blindness in many countries. Diabetic maculopathy is a condition that affects the macular region of the eye. It is considered to be critical as it affects the central vision of the eye. This paper proposes a new approach for the classification of diabetic maculopathy considering fundus fluorescein angiogram (FFA) images of healthy and clinically significant macular edema (CSME) subjects and by analyzing the size of the foveal avascular zone (FAZ) enlargement. This proposed algorithm incorporates geometric and texture measures for classifying FFA images. Supervised classifiers like classification and regression trees (CART) and random forests tree performed well with an accuracy of 96.96% for a five-fold cross-validation. This proposed method is the first step towards developing a full-fledged system for CSME detection using texture and geometry based methods.