Research On Computer Aided Diagnosis For Cerebral Aneurysm Based On DSA Images

With the rapid development of computer technology and the successful application of medical imaging in clinical diagnosis,the computer-aided diagnostic technique is growing fast.It is still to diagnose cerebral aneurysm according to observing angiography images by clinician,whose disadvantage is poor accuracy and objectivity.At present,computer-aided diagnosis for cerebral aneurysm technology is still at research stage,so there is no clinical application.This paper summarizes the current widely used computer aided diagnosis systems in clinical diagnosis,the major computer aided techniques and methods and presents a set of computer aided diagnosis for cerebral aneurysm scheme,including locating the suspected cerebral aneurysm lesion areas and statistical classification based on characteristics of cerebral aneurysm.Firstly,this paper does research on basic knowledge such as cerebral aneurysm and DSA imaging principle and so on;analyzes the processes and methods of cerebral aneurysm diagnosis by clinicians.On this basis,this paper presents the suspected lesion areas locating method based on the skeleton characteristics,which can be used to target the location of suspected lension area,and provide important details of DSA image to doctors, whose meaning is to avoid neglecting details of DSA images due to doctors themselves, and assist doctors in cerebral aneurysms diagnosis.At the same time,this paper analyzes the characteristics of cerebral blood vessels DSA images,and does fast boundary tracking according to the direction of memory,obtains the geometry information of border,and extracts the characteristic shape of cerebral aneurysm.Secondly,this paper presents the identification method of cerebral aneurysm based on statistical classification algorithm,which uses RBF neural network,support vector machine(SVM) algorithm and Bayesian algorithm to classify the shape features of cerebral aneurysm;analyzes and compares the results of the three classification algorithms.This method provides aided diagnostic information to doctors and reaches the aim of aided diagnosis. Finally,this paper summarizes the tasks have been done in project for graduation and puts forward further research suppositions.