Biomedical Application of the Evolutionary Higher-Order Spectrum

Abstract

Most of the biomedical signals are considered non-stationary since the human behavior depends on time. The ECG signal is one of the most important signals in cardiogram analysis. Although it provides a valuable basis for the clinical diagnosis and treatment of several diseases, it can be easily affected by various interferences caused by the power of magnetic field, patient respiratory motion or contraction. The overlapping interference affects the quality of the ECG waveform, leading to a false detection and recognition of wave groups. Therefore, the elimination of the interference of the ECG signal and the subsequent wave group identification has been a hot research topic. Since the ECG signal is not considered a stationary signal, neither the regular power spectrum nor the bispectrum can handle this problem because they do not reflect the time variation of the process characteristics. With the recent introduction of the evolutionary higher- order spectrum (EHOS) in digital signal processing, an approach for analyzing the ECG signal is proposed. The work in this paper is focusing on the reduction of the noise interferences of the ECG signal using the EHOS. This approach exploits the fact that the EHOS contains information regarding both the phase and the magnitude of the signal. Also, we will show that if the ECG signal is corrupted by stationary/non-stationary noise with symmetric distribution, the noise can be eliminated using the properties of the EHOS. Some simulation is declared to show the effectiveness of the proposed method.