Key Technologies Of Computer-aided Diagnosis Of Congenital Heart Disease

In recent years, the congenital heart disease attack continues to increase, thehabits and customs is changing, the heart disease becomes a great aspect which affectsthe health and happiness of people’s life. This paper focuses on two importanttechnologies of diagnosing. One is virtual endoscopy and the other is identification ofhinge point of Mitral.Virtual endoscopy is based on the3D rendering.3D texture mapping andray-casting algorithm are two main volume rendering methods. This paper focuses onthe displaying and travelling in the chamber of heart. It gives solutions to the selectionof view plain, sampling and lighting effect of virtual endoscopy. The high noisefeature and complicated structure of cardiac is the bottleneck of virtual endoscopy. Tohandle this problem, this paper gives the methods to suppress noise and displaying thecomplicated structure. This platform brings a lot of benefit for doctors to locatediseases. Cardiac virtual endoscopy is based on CT images.The hinge point of Mitral is a important point in the total heart structure which isalmost stationary in the whole cardiac cycle. This feature bring greate convenience toregistration of multimodal cardiac image and segmentation of cardiac structures. Wewill see four black chambers and some white tissue in ultrasound image which presentthe fix structure of heart. According to these features, the local context feature isdesigned to identify the hinge point of mitral. These structural features are helpful todesign a feature for the Mitral. The local context feature is calculated by sampling the8neighbor direction of a pixel. This paper adopts the additive kernel SVM as theclassifier to improve the classification. A weighted density field is designed toimprove the accuracy of the SVM classifier results. It can identify the density of theclassifier results and get the most two point sets as the approximation positions ofhinge points of mitral. At last, KMeans will reclassify the results of SVM in a specificdistance to get the accurate position of hinge point of mitral. These methods greatlyimprove the results of identification.This paper does a lot of research on two key technologies of cardiac image. These researches make great contribution to the diagnosing of heart disease. In theexperiments, we analyze the effect of the two aspects.