Research On The Algorithm Of ECG Feature Detection

The cardiovascular disease is one of the main diseases endangering human's life. ECG signal (electrocardiogram, ECG )is a synthetic reflection of the heart electricity on body surface. It has important significance to the diagnosis of heart disease by clinical ECG examination. But the ECG signal takes on complicate forms affected by the physiological states of human. Simultaneously, personal difference also makes ECG different in thousands of ways. Besides that, the random disturbance from the measure system may couple ECG to deteriorate the waveform, which makes the ECG detection more difficult. Hence, the studies on ECG processing and analyzing are of great theoritical significance and application value. Based on the research and analysis of the previous work, we made a further research on the ECG feature points detection.A detailed description of the generation mechanism of the ECG signal is firstly presented in this paper. Then each individual waveform and segment in ECG signal and its physiological significance are introduced. After the research and analysis of ECG signal detection algorithm, the problems existed in the present detection method are pointed out, which lays a foundation for the establishment of a new algorithm for the ECG signal detection.The scale selection of the traditonal dyadic wavelet method for the ECG detection is limited to the integer power of 2, which made the scale selected for the ECG feature detection is not the favorable. But the choice of scale in the continuous wavelet transform (CWT) is random, which can overcome the defect in the dyadic wavelet method. Utilizing the scale selection advantage of the CWT, we systematically studied the CWT in the application to the ECG feature detection and put forward a complete and dependable algorithm for the ECG signal feature detection based on CWT. The proposed CWT-based method has been validaeted through experiments on the MIT-BIH arrhythmia database. The simulations show that a QRS detection rate of 99.6%, a sensitivity of 99.74% and a positive prediction of 99.86 % was achieved, which proved the effectiveness of the algorithm. The wavelet method has the choice problem of mother wavelet and the decomposition scales determination problem, which means that the result of wavelet transform would be the signal under a fixed scale. From this point wavelet analysis for the ECG detection is non-adaptive. The empirical mode decomposition (EMD) has the major advantage, that is, the basis functions are derived from the signal itself. Hence, the analysis is adaptive and can overcome the choice problem of basis function and scale determination in the wavelet method. After the research and analysis of the EMD in the application to the ECG signal detection, an ECG feature detection algorithm based on the EMD is proposed. The proposed EMD-based method was validated through experiments on the MIT-BIH arrhythmia database and a QRS detection rate of 99.34%, a sensitivity of 99.77% and a positive prediction of 99.56 % was achieved. The experimental result indicated the application of EMD theory in the ECG signal detection and analysis area is feasible, and thus has laid a solid reliable foundation for the following research.