The Research Of ECG Signals Quality Assessment Approach

Cardiovascular disease is one of the most leading cause of death in the world according to the World Health Organization. The study of cardiovascular disease has became a very important topic in the academic for a long time. ECG signal is one of the important basis to the diagnosis of cardiovascular disease and evaluating whether the cardiac function is in a good condition as the comprehensive performance of cardiac electrical activity on the surface. Thus, it is one of the most important methods to diagnose the cardiovascular disease by examining and analyzing the ECG signal of patient. However, the acquisition of ECG signal is are highly susceptible to the influence of some interference factors such as the acquisition environment, leading to the quality problem appears in collected signals and deeply influenced the analysis and diagnosis of ECG signals. Therefore, it has a very important scientific value and urgent realistic demand to assess the quality of ECG signal before the analysis and diagnosis of ECG signal.In this paper, the main work was studied from the following several aspects:Firstly, the main interference sources affecting the ECG signal was analyzed. The main interference factors affecting the signal quality was analyzed in the process of collecting signal respectively from two aspects of physical and technical according to the basic physical characteristics, the signal acquisition hardware equipment and acquisition environment requirement of ECG signal.Secondly, the signal was preprocessed. The ECG signal in the database was preprocessed according to the public ECG signal quality criteria of the Physionet challenge2011.12ECG signal collecting channels of the ECG data in the database of Challenge2011Training Set A were detected to find out whether one or several of them couldn’t collect signal, if yes, the signal quality was directly determined to be unacceptable in order to save time in the signal processing.Thirdly, the ECG signal quality indexs were extracted. In this paper, The ECG signal is divided into six different frequencies, respectively is0-1Hz,1-5Hz,5-15Hz,15-50Hz,50-100Hz and100-250Hz, according to the physiological significances of ECG signal frequency range, the main energy frequency range and frequency range of the QRS complex energy. The signal energy ratio, signal energy entropy and the kurtosis value of ECG signal in the six different bands were defined as the ECG signal quality indexs. Then the quality of ECG signal in the database was assessed based on the signal quality indexs.Fourthly, the quality of ECG signal in the database of Challenge2011Training Set A was assessed respectively using threshold method and support vector machine (SVM) classification method. In the threshold value method, a threshold value was selected for each quality index according to the499ECG signal index values in the training set. ECG signal quality was determinedd as acceptable if the values of the quality indexs meet the requirements of all the thresholds. The method of support vector machine (SVM) classification, was about to select the best parameters according to the ECG signal quality indexs in the training set, train the support vector machine according to the selected parameters, assess the quality of the ECG signal in the test set according to the trained support vector machine.Lastly, the experimental results were compared between our method and other ECG signal quality assessment methods, the respective advantages and disadvantages were analyzed. The future research direction of our project was prospected based on the research and papers in present.