Research Of ECG QRS Wave Detection And Classification

Electrocardiogram (ECG, Electrocardiography) is an important basis of a doctor's pathological analysis to the basic functions of the heart, therefore, the ECG signal analysis, identification, classification has important significance.Firstly, Current research status and common testing, analysis and classification of ECG are briefly introduced and analyzed in this thesis. Then three new methods are presented. For QRS wave detection, this thesis proposed gradient-based adaptive threshold detection algorithm on the basis of R wave local gradient features and then, the algorithm was amended. The recognition rate achieves good results. After the detection of R wave, Q-wave and S wave detection was made. In terms of function approximation and feature extraction, the approximation abilities of a variety of common curve fitting functions to approach QRS wave which is a kind of special form wave is analyzed in detailed in this thesis. And then, a function which could appropriately approach the nonlinear QRS wave characteristics is found later. On classification, this thesis presents classification algorithm based on similarity threshold according to Hilbert similarity theory, and make a classification test to QRS waves directly. Then according to this algorithm, this thesis makes the necessary adjustment to the feature weights extracted by function fitting to optimize the extraction characteristics and achieves a better classification results. The classification results are practical. K-Means clustering algorithm is improved according to the similarity theory. This thesis combines the optimized characteristics with the classification algorithm to classify the QRS waves achieving good classification results.ECG classification system, which integrated QRS detection, feature extraction, data classification, results of verification and other functions, is a recognition system to assist the thesis research in the various stages of the process. In this thesis, a set of ECG data analysis processes explored based on the system makes the research have some practical significance.