The Research Of ECG Signal Myoelectricity Interference Removing And Feature Point Detection Algorithm

Heart disease does serious harm to human’s health, and even life at all times. So the research of automatic analysis system to Electrocardiograph(ECG) signal has high clinical value and it is the domestic and it is also the focus of domestic and foreign scholars. The real inspected ECG signal is easily influenceed by all kinds of inteferences signals, which futher influences the analysis and dispose of signals. This paper mainly studies the myoelectricity interference removal and the feature point detection algorithm.ECG signal is a weaker signal and the frequency is low, so it’s easily interfered by all kinds of noise in its acquisition. However, the de-noising processing for ECG signal is a prerequisite to detect the signal feature point, and the de-noising result has a direct influence on the detection result. Besides, the feature point of ECG signal contains very important physiological information, which has immeasurable academic value and practical significance.Firstly, this paper summarizes and analyzes the commonsources and types of interferences to ECG signal, mainly including baseline wander, myoelectricity interference and power line interference. The study simply introduces the wavelet transform theory and uses wavelet threshold de-noising method to eliminate baseline wander and power line interference. According to the analysis of the types of interferences, the difficulty of the research is the myoelectricity interference removal. Due to the less study on the ECG signal myoelectricity interference in the recent years, a method has been put forward by combining Empirical Mode Decomposition(EMD) and Principal Component Analysis(PCA) to remove the myoelectricity interference through the deep research in the myoelectricity interference of frequency distributing. Meanwhile, the proposed method is evaluated over MIT-BIH Arrhythmia database in terms of visual inspection and qualitatively by Signal Noise Ratio(SNR) and Mean Square Error(MSE). The results shows that the algorithm mentioned above is better than wavelet de-noising algorithm and the EMD de-noising algorithm as a whole. The problems that the signals are easily unstable and useful information may be lost have also been solved.We can obtain clean ECG signal with de-noising approches.Secondly, the enhanced signal is detected, with the wavelet tansform being in vogue at present. The core idea is looking for wavelet tansformed mould extreme value in a certain scale, and the zero-crossing of the mould extreme value accords with ECG signal feature point. From the common knowledge of ECG signal, we can know that the amplitude and slope of R-wave in QRS complex is the biggest in some areas. This article has presented a regional threshold matching algorithm to detect the feature point of waveform. The process using quadratic spline function and atrous algorithm calculate the mould exetrem value of ECG signal of their wavelet tansform are expounded. Then define the search area from positive maximum threshold. We use the positive maximum threshold as a starting point, the specific parts as the search space, and search the negative maximum threshold from the starting point to left. The value passing zero between the positive maximum threshold and the negative maximum threshold is the corresponding point of R wave. Then we detect the Q wave and S wave and the begin and the end of QRS wave based on the location of R wave. This algorithm can maintain the sampling signal integrity and be accurately positioning the QRS wave. The experimental results show that the algorithm can detect QRS wave accurately and reduce the computation complexing to a certain extent and be real-time. The amplitude of P wave and T wave is lower, that of the frequency low, but they are very important for cardiovascular diagnosis. So we made the QRS complex after being accurately positioning the QRS wave. Then we use regional threshold matching algorithm to detect the P wave during the course of quadratic spline wavelet transform, and combined with circular arc approximation method to be accurately positioning the T wave. The simulative test results show that the algorithm in this paper can reduce the false rate of sampling signal and increase sensitiveness and accuracy.At last, the paper hassummarized the main contents of the research that include the combinating the EMD algorithm with PCA technology to remove myoelectricity interference and regional threshold method to detect ECG signal feature points. The principal component weight selection threshold optimized and construct new wavelet basis function that need to be studied further.