Study Of Left Ventricular Function Evaluation And Congenital Heart Disease Intelligent Auxiliary Diagnosis Based On4D Images

Congenital heart disease is the most common heart disease in children andcaused by cardiovascular dysplasia in fetal period. There are various imagingmodalities for congenital heart disease diagnosis, such as echocardiography, magneticresonance imaging, and computerized tomography. Doctor diagnosis congenital heartdisease according to the imaging, on this basis, establishing different pathologicaldefects of quantitative indicators can reduce the excessive dependence on experienceof medical personnel in the diagnosis. The evaluation of left heart function plays animportant role for the preliminary diagnosis, the choice of treatment, and prognosisanalysis of congenital heart disease. The traditional evaluating parameters are greatlyinfluenced by individual factors, and have low sensitivity. This paper proposes a newmethod based on the shape of left ventricular time-volume curve change rule toevaluate left ventricular function and diagnosis congenital heart disease, and built athree-dimensional medical image processing platform for intelligent auxiliarydiagnosis of congenital heart disease based on the four dimensions heart images.According to the left ventricular volume change rules with time in a cardiaccycle to evaluate left ventricular function, and then prove its effectiveness. Studyfour-dimensional cardiac CT images in the21phases of patients with complexcongenital heart disease tetralogy of fallot and healthy subjects in the R-R interval ina cardiac cycle. A3D automatic anatomy segmentation method based on registrationto segment the images and extract LV silhouettes. Using pixel spatial resolutionwithin silhouettes to calculate the left ventricular volume in21phases, and draw intothe left ventricular time-volume curve. The research shown that left ventriculartime-volume curve of all healthy subjects have the same shape, but curves of tetralogyof fallot group are obvious different form healthy group, so we put forward themethod according to the shape of left ventricular time-volume curve change rules toevaluate left ventricular function.Based on AR model parameter combined with distance measurement for thecurve similarity clustering to realize the automatic clustering recognition of tetralogy of fallot based on left ventricular time-volume curve, and find a feasible method forintelligent auxiliary diagnosis of congenital heart disease. After linear normalizationtime of left ventricular volume curve, templates and samples are used to AR model,then using the Euclidean distance of AR model parameters to measure the similarityof curve, again according to the size of the similarity of curve to cluster, giving theresult of diagnosis at last. Study and compare with Euclidean distance, Hausdorffdistance, and dynamic time warp algorithm to prove the feasibility of our method.This paper also built a three-dimensional medical image processing platformwith VTK combination of MFC technology. The major function of this medical imageprocessing system includes four-dimensional cardiac CT images registrationsegmentation, three-dimensional reconstruction of the cardiac CT images, automaticevaluation of left ventricular function, congenital heart disease of intelligent auxiliarydiagnosis, and others. This is platform designed to provide a specificthree-dimensional medical image processing system for cardiac function evaluationand computer aided diagnosis of the congenital heart disease.