ECG QRS Complex Morphology Analysis Based On Feature Extraction

Electrocardiography (ECG) records a heart electric activity and is an electric potential changement figure of heart beat. ECG has a very important meaning in diagnosis of various cardiovascular diseases such as coronary artery disease, myocardial ischemia,injury and infarction.This thesis presents the current domestic and overseas ECG research state and common ECG analysis methods at first. And then the author describes the research results in three areas including the QRS complex detection, feature extraction and the classification based on feature extraction. In detecting and separating QRS complex, the author presents a combination method which combines technology by time-frequency wavelet, frequency domain filtering and the time-domain morphological analysis and verifies its feasibility. In the QRS complex feature extraction, the author presents a curve fitting idea using nonlinear least squares algorithm for QRS complex feature extraction. The feature of the algorithm is using several special function prototypes which are similar to the QRS complex for fitting. By using this algorithm, the author has done QRS complex exaction and classification experiments to MIT-BIN (arrhythmia database) database and gets good experimental results.The author has developed the QRS complex classification system based on feature extraction as an experimental platform. All the algorithms in this thesis are verified by experiments and have realistic significance.