Optimization of Human ECG Signal Acquisition Based on Wavelet Neural Network and Genetic Algorithm

—Human ECG signal is weak, the signal-to-noise ratio is small, the collected ECG signal amplification and filtering are often accompanied by interference, the interference from the body's own interference such as EMG interference, but also from outside interference such as 50Hz power frequency interference[2]. These disturbances degrade the signal-to-noise ratio of the system, and can even overwhelm the useful weak ECG signal, especially 50Hz power frequency interference. If not to eliminate to a very small extent, it will affect the subsequent signal processing, diagnosis, recognition accuracy. In addition the ECG signal also includes the following two kinds of interference: First, baseline drift [3]. Usually caused by human respiration and excitement of the heart and legs, the body's movement will lead to a certain amount of baseline drift, the frequency is lower than 1Hz, the performance of the slow changes in the curve; Second, EMG interference, which is caused by human body tremor, its frequency range is very wide. Usually between 5 ~ 2kHz[1], the performance of the rapid changes in irregular waveforms. In addition, due to the test sex, age and skin conductivity, so there are differences and the impact of equipment, system acquisition of different people's ECG signal is also very different. Therefore, in order to eliminate these interference and obtain an effective ECG signal, this paper presents a genetic neural network based on human heart ECG parameters acquisition system. By combining the nonlinear approximation ability of wavelet function with the self-learning characteristic of neural network, the output accuracy of sensor can be effectively improved, the effect of non-target parameter on detection result can be eliminated, and the detection accuracy of human physiological parameters can be improved. Through the experimental test, the ECG parameters of the device are accurate and reliable data has a good practical value.

[1]  J. Kastrup,et al.  Algorithm for the automatic computation of the modified Anderson-Wilkins acuteness score of ischemia from the pre-hospital ECG in ST-segment elevation myocardial infarction. , 2017, Journal of electrocardiology.

[2]  M. Tysler,et al.  Modeling and visualization of the activation wavefront propagation to improve understanding the QRS complex changes indicating left ventricular hypertrophy. , 2016, Journal of electrocardiology.

[3]  A. Baranchuk,et al.  New methodologies for measuring Brugada ECG patterns cannot differentiate the ECG pattern of Brugada syndrome from Brugada phenocopy. , 2016, Journal of electrocardiology.

[4]  Adrian Baranchuk,et al.  NEW METHODOLOGIES FOR MEASURING BRUGADA ECG PATTERNS CANNOT DIFFERENTIATE THE ECG PATTERN OF BRUGADA SYNDROME FROM BRUGADA PHENOCOPY , 2015 .

[5]  Kholkhal Mourad,et al.  Efficient automatic detection of QRS complexes in ECG signal based on reverse biorthogonal wavelet decomposition and nonlinear filtering , 2016 .

[6]  Jong-Myon Kim,et al.  Adaptive ECG denoising using genetic algorithm-based thresholding and ensemble empirical mode decomposition , 2016, Inf. Sci..

[7]  M. D. Uplane,et al.  High Frequency Electromyogram Noise Removal from Electrocardiogram Using FIR Low Pass Filter Based on FPGA , 2016 .

[8]  Sandro Fioretti,et al.  Segmented beat modulation method for electrocardiogram estimation from noisy recordings. , 2016, Medical engineering & physics.

[9]  Jaspal Singh,et al.  A method of REM-NREM sleep distinction using ECG signal for unobtrusive personal monitoring , 2016, Comput. Biol. Medicine.

[10]  Giovanni Danese,et al.  Custom FPGA processing for real-time fetal ECG extraction and identification , 2017, Comput. Biol. Medicine.

[11]  Yajun Ha,et al.  An energy-efficient system on a programmable chip platform for cloud applications , 2017, J. Syst. Archit..