Classification of Macula Edema Using Discrete Wavelet Transform Features

Macula edema occurs when there is fluid appearing around the macula. The macula is responsible for the central vision and these fluids will block the central vision. The severity of this condition depends on distance between the fluid and macula. Macula edema may be caused due to either cystoid or diabetes, where diabetes is the more common cause. In this paper, images are classified into normal, non-clinically significant macula edema (NCSME) and clinically significant macula edema (CSME) using features extracted from the discrete wavelet transform (DWT). The clinically significant features are input into six classifiers to select the best classifier. In this work, we have obtained the highest average accuracy of 80%, with a sensitivity of 93.5% and specificity of 87% using support vector machine (SVM) with a kernel of polynomial 2.