Analysis Electroencephalogram Signals Using Denoising and Time-Frequency Techniques

Abstract

The electroencephalogram (EEG) is a test that determines brain activity. The existence of artifacts in EEG can naturally decrease the smoothness of the analysis of the biomedical signal. EEG disturbed by noises during encephalogram recordings is one of the problems that the experts have to investigate for finding solutions to remove these artifacts. To obtain an accurate EEG signal; the improved complete ensemble empirical mode decomposition with adaptive noise (ICEEMDAN) is used and compared with and its old versions. The non-stationarity and non-linearity of the EEG signal cannot provide complete information when using the traditional method (Temporal and frequency domains). Among the objectives is the comparative analysis of time-frequency distributions applied on EEG signals. The denoising and time-frequency methods used in this study are tested on healthy and abnormal EEG signals which are disturbed by natural and artificial noises. The comparative study in this work shows the effectiveness of the combination of the ICEEMDAN and Periodogram methods that are suitable for denoising and analyzing the EEG signal.