基于块稀疏贝叶斯学习算法的心电数据重构

Electrocardiograph Reconstruction Based on Block Sparse Bayesian Learning Algorithm

压缩感知(CS)技术在心电信号上的应用具有低成本、低功耗等优势,但传统的CS算法重构心电信号质量并不理想。本文介绍了一种基于信号块结构内相关性的块稀疏贝叶斯学习(BSBL)CS算法;并对MIT-BIH数据库中心电数据进行实验,结果显示其均方根误差远低于传统CS算法,表明该算法能够高质量重构心电信号。BSBL算法在心电数据上的应用有效降低了对数据的采样频率,从而缓解存储压力并降低功耗。

Compressed sensing（CS） has the advantages of low cost and low power consumption in the application of ECG signal, but the quality of the existing algorithm to reconstruct the non-sparse ECG signal is not ideal. This paper introduces an ECG data reconstruction method based on the theory of the compressed sensing. Block spare bayesian learning（BSBL） algorithm combined with intra-block correlation is used to efficiently reconstruct ECG data from MIT-BIH database. The simulation results show that the method can accurately reconstruct ECG signal, compared to traditional CS algorithm with lower root mean square error（RMSE）, and reconstruction quality is improved significantly. BSBL algorithm can efficiently reconstruct the non-sparse ECG signal, reducing the use of sensors and power consumption.