心音信号时频分析方法研究

目的:研究分析适用于心音信号的时频分析方法,该方法具有较高的时频分辨率,能够充分表现不同心音信号的时频特性,实现心脏病辅助诊断.方法:对心音信号进行相关预处理后采用维格纳分布、短时傅里叶变换和S变换3种方法分别对心音信号进行分析.结果:维格纳分布法具有良好的时频聚集性,且分辨率高,但含有交叉干扰项;短时傅里叶变换的时频聚集性差,且分辨率低;S变换的方法时频分辨率非常高,时频聚集性良好.结论:S变换具有宽度可变的高斯窗,频率分辨率高,可以满足心音信号时频分析对频率分辨率的要求

Methods for Time-frequency analysis of heart sound

Objective: To study a suitable method for time-frequency analysis of heart sound, which has good time-frequency resolution to display the features of different heart sounds and assists the diagnosis of heart diseases. Methods: Wigner-Ville distribution, short-time Fourier transform and S transform were used to analyze the heart sound after the pre-processing of heart sound. Results: Wigner-Ville distribution had good time-frequency clustering and high resolution, but it contained cross interference items; short-time Fourier transform had bad time-frequency clustering and low resolution; S transform had high time-frequency resolution and good time-frequency clustering. Conclusion: S transform can meet the requirements for the analysis of heart sound because of its Gaussian window, the width of which can change and high time-frequency resolution.