Study On Compression Reconstruction Algorithm For ECG Signals Based On Approximate Message Passing

In recent years,the prevalence of cardiovascular disease in China is on a continuous rise due to unhealthy lifestyle habits and an aging population,and cardiovascular disease has become a major public health problem that we urgently need to address.As an important clinical medical examination,the large amount of data generated by long-term ECG monitoring puts great pressure on transmission.Compressed sensing technology breaks the limitation of sampling frequency in Nyquist sampling theorem,which not only reduces the pressure at the signal acquisition end,but also reduces the amount of data transmission.The key problem of compressed sensing technology is how to ensure the high precision reconstruction of the signal,so the reconstruction algorithm has been a hot issue in the research field of compressed sensing.This paper focuses on completing the research of ECG signal reconstruction and denoising based on approximate message passing.The main contents include:(1)To solve the problem of high complexity of the traditional ECG signal compression-aware reconstruction algorithm,the GAMP-SBL ECG signal reconstruction algorithm is proposed to reduce the complexity of the algorithm and the reconstruction time.In this paper,on the framework of sparse Bayesian learning,the prior information of ECG signal is extracted,and the generalized approximate message passing algorithm is used to replace the matrix inversion of E-step in the EM algorithm,and the scalar function is solved by the Max-Sum algorithm,and the reconstructed ECG signal is obtained by convergence in iterations.Simulation experiments on the MITBIH arrhythmia database show that the GAMP-SBL algorithm has a shorter reconstruction time compared with the conventional compressed sensory reconstruction algorithm in obtaining ECG signals with the same reconstruction accuracy.(2)Reconstruction accuracy is an important index of ECG signal compressionaware reconstruction.from improving the reconstruction accuracy of ECG signal,a vector approximation message passing reconstruction based on denoising(D-VAMP)algorithm is proposed.the D-AMP algorithm can achieve the approximation of the original message by selecting the appropriate denoising function.in this paper,the VAMP algorithm is used for ECG signal reconstruction instead of the AMP algorithm to reduce the complexity of the algorithm.Since the noise in the ECG signal behaves similarly to Gaussian noise,the NLM denoising function suitable for removing Gaussian noise is selected to obtain the D-VAMP algorithm based on ECG signal denoising.Experiments show that the algorithm can effectively reduce the reconstruction error of ECG signal and retain more effective information of ECG signal,and the PRD value is 8.6012 at the compression rate of 81%,which meets the diagnostic requirements of ECG signal.(3)A convolutional sparse coding-based denoising algorithm for ECG signals is proposed.The noises interspersed in the ECG signal sampling can mask its own feature information and affect the subsequent feature analysis identification and diagnosis.A convolutional sparse coding dictionary is used to learn a convolutional dictionary with ECG signal features,and the noisy ECG signal is processed by convolutional sparse coding and reconstructed to obtain a denoised ECG signal using this dictionary.The experiments show that compared with the traditional ECG signal denoising methods,the convolutional sparse coding-based ECG signal denoising retains the morphological features of the ECG signal to the maximum extent while denoising,and the SNR reaches 21.6828,which verifies that the algorithm has a better denoising effect on the ECG signal with high signal-to-noise ratio.