잡음 대응성을 강화한 심음도 기반 심장 박동 감시 알고리즘 개발

Abstract

Accurate and reliable real-time monitoring of heart beat status is important for out-hospital patients with the risk of cardiac arrest. In this paper, I proposed an enhanced phonocardiogram (PCG) -based noise-robust heart beat monitoring algorithm and evaluated the performance of the proposed algorithm using six subjects. A commercialized electronic stethoscope was utilized to measure phonocardiogram signal, and additional electrocardiogram (ECG) signal was measured for the purpose of reference. The proposed algorithm was composed of three main sub-parts: attach/detach monitor, artifact classifier, and heart beat calculator. The attach/detach monitor decides whether the stethoscope is normally attached on the body surface or not. The artifact classifier decides the type of artifact among rest, voice, and walk. And the heart beat calculator processes the input signal to reduce the artifact components and then calculates the value of heart rate. In experiments for single-situation, difference between heart rates from ECG and the conventional PCG-based algorithm was 2.73 ± 1.18 BPM in rest, 18.61 ± 15.86 BPM in voice, and 6.53 ± 7.37 BPM in walk, respectively; in contrast difference between heart rates from ECG and the proposed algorithm was 1.58 ± 0.74 BPM in rest, 3.76 ± 4.29 BPM in voice, and 2.16 ± 1.58 BPM in walk, respectively. In experiments for serial-situations, difference between heart rates from ECG and the conventional PCG-based algorithm was 3.48 ± 2.57 BPM in rest, 7.31 ± 6.68 BPM in voice, and 7.17 ± 7.00 BPM in walk, respectively; in contrast difference between heart rates from ECG and the proposed algorithm was 2.26 ± 0.76 BPM in rest, 2.93 ± 1.86 BPM in voice, and 2.50 ± 1.13 BPM in walk, respectively. Based on these experimental results, I propose that the proposed algorithm have a potential for accurate and reliable real-time heart beat monitoring purpose.