Diabetic Macular Edema Detection using Graph Theory and Retinal Layer Features

Diabetic Mellitus is a disease which can affect the functioning of different organs in the body such as heart, kidney, retina etc. When this disease affects the eye it leads to a condition called Diabetic Retinopathy and this disease in turn may lead to a disease called Diabetic Macular Edema (DME). If DME is left untreated it may lead to vision loss. In this work, a method of automated detection of DME from retinal images has been proposed. Retinal layer in the Optical Coherence Tomography (OCT) images are first segmented using graph based method. Then, a set of 30 features are extracted from the retina and this has been given to different forms of Support Vector Machine (SVM) for the classification purpose. SVM with polynomial order 2 gives the best result with 96.66% accuracy. Keywords— Diabetic Macular Edema, Optical Coherence Tomography, Graph Theory, Retinal features, Support Vector Machine.