基于L2范数和L0范数方法的心电逆问题研究-计算机科学与技术专业论文.docxVIP

关键词：ECG逆问题；平滑LO范数；L2范数 关键词：ECG逆问题；平滑LO范数；L2范数 万方数据 AbstractCardiac Abstract Cardiac is one of the most important organs of human，and its electrical activity is a very important physiological phenomenon of human’S life activities．The early detection and localization of some specific electrical signals of heart diseases are highly necessary．However，the existing approaches are invasive or mainly depend on doctor’S experience，they ale not accurate and cause great pain to the patients．A non-invasive while quantitative approach for the precise localization of focus is badly needed now．Thus，ECG inverse problem has drawn much attention． The study of ECG inverse problem is to solve the problem that gets the electrical information of cardiac based on the body surface potential information and the physical and geometric properties of the body volume．ECG inverse problem is to study the method of invasively and quantitatively reconstruct cardiac electrical signals The ill-posedness of ECG inverse problem brings a huge challenge to people，that a slight noise in body surface potential data leads to a huge deviation in reconstructing results．In this study，we first used a low computational complexity and validate TSVD(truncated singular value decomposition)method which is based L2-nolTn，and determine the regularization parameter by L—curve．We did experiments on dataset of isolated canine heart，reconstructed and visualized the EPs．The results show that most of the features have been reconstructed，the proposed L2-n01Tfl method is a valid approach to reconstruct EPs．While the penalty function of L2一norlTl causes a smoothness to the solution，decreasing the accuracy of detecting abnormal tissue and locating focus．The smoothing reduces the accuracy of locating the diseased location． The non·-differentiability of penalty function of L 1·-norm is making the computational complexity considerably big．In Our study，a smoothed L0-nolm regulation method has been used to solve the LO—nolTn constr