Research on Parallel Compressive Sensing and Application of Multi-Channel Synchronous Acquisition of Heart Sound Signals

Abstract

The rapid development of wearable biosensors based on the Internet of Things provides a new solution for heart disease analysis and prediction of sudden death from heart failure. However, how to deal with a large amount of data generated by heart sound signals is an urgent problem to be solved. Based on the characteristics of heart sound signals, we propose a parallel compressive sensing model for the multichannel synchronous acquisition of heart sound signals. Furthermore, we provide a series of experiments to assess the performance of the model. The experimental results demonstrate that the reconstruction speed of the proposed model is 9-10 times faster than that of the block sparse Bayesian learning algorithm and the orthogonal matching pursuit algorithm, and the reconstruction effect is better. Meanwhile, the proposed model can effectively reconstruct the normal heart sound signal and abnormal heart sound signal of fourchannel synchronous acquisition. Therefore, the proposed model is feasibility. INDEX TERMS Heart sound signals, compressive sensing, parallel signals, monitoring, L 1 norm, block sparse Bayesian learning. I. INTRODUCTION

Comparative Study of Compressed Sensing for Heart Sound Acquisition in Wireless Body Sensor Networks

“…The systolic period of the heart lasts from S1 to S2, and the diastolic period lasts from S2 to S1. Normally, diastole is slightly larger than systole [27]. When the heart beats too fast, diastolic periods shorten, and further make the systolic and diastolic periods identical.…”

“…According to the different stethoscope area, the intensity of HS will also be different. The S2 signal of aortic area is strong, and the S1 signal of bicuspid valve and tricuspid valve stethoscope area is strong [27]. For the waveform of a group of HS, the three values S1/S2, W1/W2 and D/S obtained by medical indicators can determine the congestive state of the heart [33].…”

“…In comparison with the extensive research on ECG and EEG acquisition, there are few previous works focusing on acquiring HS using the CS technique [26], [27].…”

“…Sejdic and Chaparro [26] had proven that CS is suitable for acquiring HS, and demonstrated that 40% samples are sufficient to reconstruct the original HS signals accurately. Cheng et al [27] proposed and verified a multi-channel CS model for HS acquisition. The achievements of [26] and [27] fully proved that the CS is suitable for the reconstruction of heart sound signal.…”

“…Cheng et al [27] proposed and verified a multi-channel CS model for HS acquisition. The achievements of [26] and [27] fully proved that the CS is suitable for the reconstruction of heart sound signal. However, extensive study of CS for HS acquisition is lacking at present.…”