心音自动分段算法研究

目的 实现心音信号的自动分段：第一心音、收缩期、第二心音、舒张期。方法 提出了一种新的无需心电参考的心音自动分段算法，首先用最优的小波阈值滤波技术对心音信号预处理，保留并突出心音信号的基本特征；然后利用希尔伯特变换提取信号包络，对包络信号采取一定策略实现心音的自动定位。结果 对临床采集得到的大量正常人心音与心脏病人心音数据对算法进行验证，尽管心音信号复杂多样，用此算法的正确分段率超过95％。结论 心音自动分段算法正确率高，鲁棒性强，为各种心脏病的进一步诊断奠定了良好的基础。

A Study on Segmentation Algorithm of Heart Sound

Objective To develop a new segmentation algorithm of heart sound without ECG signal as reference. Method The heart sound signal was preprocessed by optimal wavelet shrinkage denoising and the characteristic part of heart sound were preserved. Then the envelope of heart sound was extracted. The heart sound signal was separated into four parts: the 1st heart sound, the systolic period, the 2nd heart sound and the diastolic period by enveloping signal with some strategies. Result The algorithm was tested using large amount of clinically normal and abnormal heart sound data. Whereas the heart sound signals were intricate, the algorithm achieved a detection rate of 95%. Conclusion The correct detection rate is very high by the segmentation algorithm and the algorithm was robust. The algorithm establishes a basis for the diagnosis of heart diseases.