An approach to classification of retinal vessels using neural network pattern recoginition

Artery/Vein classification of retinal vessels is a key phase in detecting automatic vascular changes. Retina is a layer, found at the back of eye ball which when damaged may leads to severe problems. Retina plays a vital role in visualization. There are two kinds of blood vessels in vascular system. They are Artery and Vein. Analysis of vascular system leads to early detection of diseases like hypertension diabetes retinopathy, cardiovascular condition, etc. Therefore, we proposed an automatic early detection of diseases by classifying Artery/Vein which resides on functionality of graph. We used Neural Network Pattern Recognition from MATLAB for classification. Graph is extracted from segmented retinal images from which the features are extracted. Morphological operations are performed on segmented retinal images. Center lines are identified using graph based approach. Then, feature extraction is performed, which is classified by neural network pattern recognition classifier in order to obtain best accuracy in classifying normal or abnormal Artery/Vein. Keywords—Retinal Vessels; Neural Network; Segmentation; Artery/Vein; Pattern Recognition.