Blink Elimination in EEG signals using Two phase Variational Mode Decomposition

The Electroencephalogram (EEG) are the data collected from human scalp to analyze functioning of the brain. These EEG signals usually contaminate with other signals like ocular artifacts (OA). There exist signal decomposition methods like wavelet decomposition, empirical mode decomposition (EMD) and Variational Mode Decomposition (VMD) to extract these artifacts. In this research work, a novel two-phase variational mode decomposition (TPVMD) is used to analyze EEG signals to extract OAs. In existing empirical mode decomposition (EMD) and multi variational empirical mode decomposition (MEMD), balancing parameter and regression is not applied. In this proposed method, Variational Mode Decomposition (VMD) is used to eliminate the OAs in two phases and in each phase, regression is applied. In this paper the EEG signal data is collected from semi-simulated Mendeley database. Blink artifacts are eliminated from corrupted EEG signal and the signal to artefact ration (SAR) for different balancing factor alpha(α). The simulation results show that the proposed method gives improved SNR and reduced difference in power spectral density (DPSD).