Ct Cardiac Image Segmentation And Left Ventricular Area Matching Technology

Cardiac image analysis is a focused topic in the field of medical image research, which is of great value in the clinical diagnosis of heart disease. Cardiac CT image segmentation and Left ventricular region matching is an important issue in cardiac image analysis. In this paper, in regard to data characteristics on cardiac CT images, we use the segmentation method of Gaussian probability simulation threshold and the feature extraction of centroid increment boundary to achieve cardiac CT images segmentation and Left ventricular region matching.In this paper, the theoretical knowledge and practical methods on medical image segmentation and feature extraction are studied and analyzed. In the process of cardiac CT image segmentation, the cardiac region showed stronger contrast characteristics of gray level due to the injected contrast agent, the density function of continuous Gaussian probability is used to threshold segmentation to achieve the segmentation of various cavity regions in cardiac CT images. This method integrates the thoughts of the histogram valley method and information entropy threshold, solves the problem which is made by the uneven distribution of contrast agent. It also improves the applicability of CT image segmentation. In the feature extraction of the left ventricular, the method of centroid vector increment matrix is used as the mathematical description for region boundary. In the method, the sampling strategy on minimum area feature points is proposed to solve cross-border phenomenon in the sampling process and construct centriod vector increment matrix accurately. Lastly, the boundary feature of left ventricular region and each connected region are extracted.Finally, in this paper, Euler distance and geometric distance are introduced into the proposed method of similarity distance calculation. This method adds the impact element of number of matrix elements for characteristic parameter of centroid matrix vector increment. The experiment results show that the description of the similarity distance reduces the similarity interval and increases the distinction between the left ventricular region and other regions.