Study On Computer Aided Diagnosis For Cardiovascular Disease Based On Medical Images

Cardiovascular disease which often arises from coronary artery stenosis is an important cause of harming human health. Currently, visual estimation by angiograms is main diagnosis method used in clinic. But this method suffers from inter-and intra-observer variability. Combining the computer and medical image processing and analysis techniques, this thesis makes a systematic research on quantitative auto-analysis of left ventricle and vessel by means of angiograms. For the auto-analysis of vessel stenosis, several image segmentation algorithms are analyzed and employed to extract vessel structure from angiogram. Then, a path-line detection method is presented based on ridge-line tracking. By interactive specified a start-point and an end-point, the lumen centerline between them can be detected automatically by the proposed method. With the path-line as a basic mode, the vessel contours are extracted using min-cost algorithm. Finally, vessel diameter function is computed which is the basis of vessel quantitative analysis. In addition, the change of left ventricle functions is important reflect of cardiovascular disease, so an automatic analysis method for left ventricle function is proposed in the end. By the computer aided measurement, the function parameters of the left ventricle can be obtained automatically.