Time-Frequency Analysis of Heart Sound Signals Based on Hilbert-Huang Transformation

Abstract

In this study, a method based on Hilbert-Huang Transformation (HHT) for time-frequency analysis of heart sound signals is presented. HHT is employed because most biomedical signals such as Electroencephalogram (EEG), Electrocardiogram (ECG) and heart sound signals are non-stationary signals. Heart sound signals recordings are often contaminated with the spike noise caused by the front-end circuits or measurement instruments in the real situations. A digital median filter is firstly employed to remove the spike noise of the heart sound signals. Then, the time series data are decomposed into several IMFs (Intrinsic Mode Function) using Empirical Mode Decomposition (EMD) algorithm. Hilbert transformation algorithm is utilized to acquire the instantaneous frequency for every IMF. Simulation results show that time-frequency domain analysis of heart sounds signals based on HHT algorithm is able to offer higher frequency resolution.