An EEG signal denoising method based on ensemble empirical mode decomposition and independent component analysis
Huimin Sun, Jun Sheng Cheng, Zheng Ma · 2018 IEEE International Conference on Cyborg and Bionic Systems (CBS) · 2018
Event-related potential (ERP) signal is very weak. The actual EEG signal is accompanied by a large amount of neurogenic noises and non-neurogenic interferences. Aimed at the issue of reconstructing the valuable ERP signal from polluted EEG signal, an efficient approach that combining ensemble empirical mode decomposition (EEMD) and independent component analysis (ICA) is proposed. Firstly, the noisy signal is decomposed by EEMD, and several interested intrinsic mode functions (IMFs) components are obtained based on correlation between the decomposed IMFs and the original signal. Then, these interested IMF components are de-noised with ICA. Finally, the desired ERP signal is reconstructed from these de-noised IMFs. Experimental results show the merit of the proposed algorithm on EEG signal de-noising.