Research Of ECG Automatic Analysis Techniques

ECG has been the main way that people understand their cardiac characteristicsand an important basis for disease diagnosis. The generation of dynamic ECG leads tomanual analysis of all the data being impossible. in order to improve the efficiency ofdiagnosis and real-time monitoring of patients, the birth of automatic ECG analysis isinevitable.ECG signals are weak, signal acquisition typically includes a variety of noiseinterference.Therefore, we must first deal with the signal de-noising, de-noising is thebasis of QRS wave detection and feature extraction, and it will directly affect theresults of the automatic diagnostic analysis.QRS wave is the most obvious part ofECG, containing a lot of important physiological information.Thus, QRS detection isan important step in the automatic analyzer, not only the positioning basis of otherwaveform but also a prerequisite feature extraction.It will affect the accuracy of theautomatic analysis and diagnosis.Based on the results of previous studies, we mainlydo further research on the QRS wave group detection technology.In denoising, this paper uses a wavelet threshold value method.The main work:1.Select the appropriate wavelet function to determine the wavelet decompositionlevel.2. Select the appropriate threshold function and threshold estimation method.3.do the simulation experiments.In this paper, we use sym8wavelet to decompositeECG wavelet, and use wavelet soft thresholding to process signal, we also outputtwo parameters-the signal of noise ratio (SNR) and minimum mean square error(MSE) to evaluate the experimental results.the results show that this method caneffectively remove main ECG noise, and has a good de-noising effect.In QRS detection, this paper presents a method for locating R peaks based onwavelet transform. The method uses a Gaussian wavelet as the wavelet function,andselects energy concentration,weak noise interference-the third layer waveletcoefficients as the research object.The main tasks:1Initialize threshold and makesure automatically threshold transformation rules.2Find the extreme points that meetthe threshold conditions, use a certain methods and optimization strategies to make pairs of extreme points correctly.3Use the extreme points position to determine therange of R peaks in the original signal, and identify the most value in this range. Thevalue of the position is the R peak position.4Use refractory periods of physiologicalprinciples to detect the position results on R.5Do simulation experiment. We doexperiment on the MIT-BIH database of some typical waveform data.The experimentresults show that the algorithm is of high accuracy positioning R peaks, which is avalidity algorithm.On the basis of locating R peaks, the paper realized the extractionof QRS Wave group width.The main work:1Determine Q wave and S wave’sroughly relative position to R wave, look for the extreme point of this range, andcorrectly make pair of these extreme points.2Use the extreme points position todetermine the range of R peaks in the original signal, and identify the most value inthis range. The value of the position is the R peak position.(Q or S peak).3Look forthe maximum change sampling point in the slope of the eight sampling points beforeS or after Q, and it is considered as the starting point for the Q-wave or the end ofS-wave, which is the starting point and the end point of the QRS wave group, thencalculate the QRS wave group width5. Do the simulation experiments. Sign thewidth of QRS wave group, Q wave crest and Swave crest. The MIT-BIH databaseportion of a typical waveform data experimental results show that this method hasgood accuracy.