Diabetic retinopathy caused by complications of diabetes, which can eventually lead to blindness. It affects up to 80% of all patients who have had diabetes for 10 years or more. Despite of these statistics, research indicates that at least 90% of new cases could be reduced if there was proper and vigilant treatment and monitoring of the eyes. The longer a person has diabetes, higher the chances of developing diabetic retinopathy .The aim of the project is to detect abnormal vessels in the optic disc of human eye and also prevent from the eye related disease by measuring the features (shape, position, orientation, brightness, contrast) and applying segmentation by replacing the values of the feature measurements the vessels are detected. The existing system uses support vector machine (SVM) to categorize each segment as normal or abnormal. The SVM is used to analyze data and recognize patterns but it cannot detect the vessels automatically, accuracy is not clear and the prediction of disease needs better knowledge. The proposed system uses neural network algorithm for training the features and prediction of disease that is accurate and faster and can find the disease based on ranking its features. .