Detection of retinal blood vessels using curvelet transform

Retinal vessel detection is an important step in diagnosing and treatment of many diseases affecting the retina. The method presented in this work helps in automated extraction of retinal vessels and aids in early detection of diseases like diabetic retinopathy. Since the curvelet transform can represent edges efficiently, the curvelet transform coefficients are modified to enhance the image. Segmentation is done by Support vector Machine which classifies each pixel as vessel or nonvessel, based on the feature vector of the pixel. The segmentation's performance is measured in terms of accuracy, sensitivity and specificity. The performance evaluation of the method is done on the publicly available DRIVE database. Better results have been obtained for segmentation after curvelet-based enhancement.