Research on removal algorithm of EOG artifacts in single-channel EEG signals based on CEEMDAN-BD

Abstract

Single-channel electroencephalography (EEG) signals are more susceptible to electro-oculography (EOG) interference, which could be attributed to the acquisition device of the single-channel. To realize EOG artifacts separation in this paper, the blind deconvolution (BD) model was investigated based on complete ensemble empirical mode decomposition with adaptive noise (CEEMDAN). The CEEMDAN method was firstly used to decompose the EEG data contained artifacts into several intrinsic mode functions (IMF). After that, the modal component used as the observed signal was provided to the BD model, which was formed by the source signal of the EEG signal and the EOG artifacts. Consequently, we successfully realized the separation of EEG signal and EOG artifacts by the constructing cost function iteratively, and our results demonstrated that the separation effect of this method on EOG artifacts is better than previous studies. Further, the correlation coefficient of real-life data after CEEMDAN-BD algorithm processing reaches 0.81. Moreover, the modal aliasing problem was solved with most of the original EEG signal components retained. In a word, this novel method provides theory and practice references for the processing of EEG signals and other physiological signals.

Keywords:

• EEG signal
• EOG artifacts
• empirical mode decomposition
• blind deconvolution model

Disclosure statement

There are no conflicts of interest to disclose.

Additional information

Funding

This work was supported in part by the Jiangsu key R & D plan (No. BE2020648, No. BE2019317), National Natural Science Foundation of China (No. 62001196), Qinglan Project in Jiangsu Province(No. KYQ19014) and Teaching Reform Project of Jiangsu University of Technology (No. 11610211943).