The Design Of ECG Singal Clustering Algorithm

Heart disease is one of the major diseases that threaten human life.Research on heart disease has been an important issue in the medical profession for a long time.Human electrocardiogram contains a rich heart rhythm and heart’s electrical conduction of physiological and pathological information as a comprehensive manifestation of cardiac electrical activity in the body surface,VCG is developed after the ECG,but it is the basis of the theory of ECG.Both of them are reflected in the ECG activity,the recording method is different,and every one has its own advantages.A 24-hour dynamic VCG is filled with massive amounts of datas in today’s information society.Especially in some cases where useful datas is only a small part.Discussing how to extract useful information and translate it into contents which is clear and easy to analyze is a practical topic in this paper.This paper mainly studies the clustering algorithm of VCG signal,discusses the principle and development status of clustering algorithm in detail,and analyzes the key technology and the difficult problem in the process of 3D VCG ring clustering research.First of all,by doing extraction process on VCG datas,a unified format of the source datas which will be easy for post-analysis can be generated.Then,the 24-hour datas is clustered based on mathematical morphology,the redundant main classes will be recognized.Clustering the remaining datas based on density at last.Chosing the area and height of QRS ring,the distance with the template of normal and ventricular premature beat,the pre-RR and late-RR duration of the QRS ring as the characteristic parameters.The final results will be arranged,the cluster center will be obtained,and the type of every beat will be marked.In order to verify the reliability of the algorithm,the paper choose the the 24-hour clinical ECG datas which are marked the type by the expert as the object of the detection algorithm.The final results show that the method achieve a satisfactory result in the overall accuracy of clustering,and it has a high detection rate for rare pathological datas.The accuracy rate about main type of heart beat has reached more than 97%,the processing time for 24h heart beat has reached 108.99s.