Research On ECG Signals Denoising Based On Optimized VMD And Improved Wavelet Threshold Algorithm

The number of people suffering from cardiovascular disease is increasing year by year in our country.Therefore,the prevention and treatment of cardiovascular diseases is greatly significance to our citizens’ health security.Electrocardiogram(ECG)signals are collected by ECG detection equipment that the human heart electrical signals’ waveform change.Doctors can diagnose the physiological condition of the human heart by observing the waveform shape and related characteristic parameters.However,the ECG signals inevitable will be interfered by a variety of different frequency noises in the process of acquisition and transmission,such as baseline wander,power line interference and motion artifacts.These noises will cause the ECG signals waveform to be distorted to varying degrees,and cause inaccurate follow-up clinical analysis and diagnosis results.Therefore,developing a powerful performance algorithm that remove the ECG signals noise accurately are essential to follow-up cardiovascular diseases’ research and diagnosis.The paper has proposed two improved ECG signals denoising algorithms by analyze and summarize the existing denoising algorithms’ defects and combining the principle of optimization algorithms.The ECG signals are selected the real ECG signals of the MIT-BIH arrhythmia database,and the noises come from the real noises of the MIT-BIH noise pressure test database and the simulated noise formed by add the gaussian white noise and sine signals to ECG signals in this paper.Through quantitative observation and comparison of the denoised ECG signals’ waveform characteristics and qualitative comparison of the denoised ECG signals’ signal to noise ratio(SNR),mean square error(MSE)and Pearson correlation coefficient(PCC)values,the denoising effect of each algorithm is comprehensively evaluated.(1)An improved variable step LMS algorithm based on fractional functionTo the problem that the fixed step size LMS algorithm and the existing variable step size LMS algorithms have poor denoising effect on ECG signals,this paper proposes an improved variable step size LMS algorithm based on fractional function.First,the step-size function of the improved variable step-size LMS algorithm is constructed by using the fractional function,and the optimal value of the algorithm parameters is obtained through theoretical and simulation analysis.Under the same conditions,the performance comparison with the fixed step-size and other variable step-size LMS algorithms verifies that the improved algorithm has faster convergence speed,lower steady-state error and smaller computational complexity.In order to demonstrate the improved algorithm’s denoising performance for noisy ECG signals,which compared to fixed step size and other variable step size LMS algorithms for multiple real noisy ECG signals under the same conditions.The experimental results show that the proposed improved algorithm can better remove the noise of ECG signals,and the qualitative and quantitative results are better than other related algorithms.(2)Optimized VMD and improved wavelet threshold algorithmIn order to solve the problem that the existing correlation algorithms have poor denoising effect on ECG signals,this paper proposes an ECG signals’ denoising algorithm that optimized VMD and improved wavelet threshold algorithm.In this paper,genetic algorithm is used to optimize the parameters of VMD algorithm,which better solves the problem of human experience setting decomposition scale k and penalty factor α and only paying attention to a single parameter cause VMD decomposition poor effect.Based on the hyperbolic tangent function,an improved adjustable threshold function with continuity,simple structure and good flexibility are constructed,meanwhile,a layered threshold that better conforms to the distribution pattern of noise in each layera is proposed,the optimal wavelet basis function and the number of decomposition layers are obtained through theoretical and simulation analysis,therefore,the improved wavelet threshold algorithm is proposed.First,GA uses envelope entropy as fitness function to iteratively optimize the search of VMD algorithm parameters [k,α] best combination value,and the ECG signals with real and analog noise are decomposed by VMD algorithm to get k IMF,and solve the PCC value and spectrum of original ECG signals’ each IMF components to comprehensively determine which IMF component with highfrequency noise and low-frequency noise.Then the IMF components of low-frequency noise is discarded,and the improved wavelet threshold algorithm is used to denoise other IMF components with smaller PCC value,the IMF components with larger PCC value and the threshold algorithm denoising IMF components are superimposed and reconstructed to obtain the denoised ECG signals.Finally,the algorithm is compared with other algorithms to the denoising effect of ECG signals with simulated and real noise,The qualitative and quantitative results show that the improved algorithm can better remove the ECG signals noise and maintain better waveform characteristics,meanwhile,the SNR,MSE and PCC values of the ECG signals after denoising have improved to varying degrees.To sum up,this paper proposes two improved ECG signals denoising algorithms,which can better remove the noise of ECG signals and maintain better waveform characteristics.It can provide theoretical basis for the subsequent research of ECG signals denoising algorithms.