Research On The Microaneurysm Detection Algorithm Using KPCA And SVM

The fast development of computer technology has driven the digital retinal image processing and analysis technology to be more mature. There are a number of diseases, particularly vascular disease, which leave tell-tale markers in the retina such as diabetes. Diabetic retinopathy is a common and severe complication of diabetes, which damages to the retina and even leads to blindness. Therefore, early detection of diabetic retinopathy is of vital importance for preventing significant vision loss. Since microaneurysms are regarded as the first clinical symptom of diabetic retinopathy, the accurate detection of microaneurysms in retinal images is a critical step for early detection of diabetic retinopathy. This research aims to propose an efficient approach for microaneurysm detection in digital fundus images through the simulation experiment.This thesis first overviews the research background and significance, introduces the research status at home and abroad. According to the basic principles of microaneurysm detection, there are three key steps for an efficient microaneurysm detection approach:the acquisition, feature extraction and feature classification of microaneurysms candidate. The acquisition of microaneurysms candidate can be achieved by using appropriate image preprocessing technology and candidate microaneurysms extraction theory.This thesis mainly researches on the feature extraction of microaneurysms candidate using principal component analysis algorithm and kernel principal component analysis algorithm.The essence of principal component analysis algorithm is to make the candidate samples in retinal microaneurysms candidate set from higher dimensional feature space to a lower one by linear mapping transformation, and complete the dimension reduction operation. Due to the ignorance of relationships among the higher order feature index in principal component analysis algorithm, the main components through dimension reduction can not fully express the information carried by the original high dimensional feature index. Therefore, kernel principal component analysis algorithm is introduced to achieve the feature dimension reduction, which principal component analysis is performed in the kernel space through appropriate kernel function. Experiments indicate that this novel extraction algorithm can effectively improve true positive rate of microaneurysm detection.Finally, the lower-dimensional features of microaneurysms candidate achieved by feature extraction are used for classification with support vector machine algorithm. There are two steps for the support vector machine algorithm. In the support vector machine algorithm, some samples of lower-dimensional microaneurysms candidate are used to train a classifer and the others for the detection of classifier effect. Besides, particle swarm optimization algorithm is applied to determine the optimal kernel function parameter σ and the optimizing penalty factor C in SVM algorithm. Experiments have shown that SVM algorithm can decrease false positive rate in microaneurysm detection.To sum up, microaneurysm detection algorithm based on KPCA and SVM in this thesis can increase detection sensitivity, and it also decrease the average false positive rate while maintain the same true positive rate. Therefore, the proposed algorithm is effective.