EOG Eye Blink Artifact Removal using Multivariate Variational Mode Decomposition and PCA

Artefacts in an electroencephalography (EEG) recording are most commonly caused by eye movement, which has a higher electrical potential than a typical EEG recording. Only by removing all ocular artefacts (OA) with little to no loss of EEG signal can a higher quality recording be obtained. This paper proposes a method based on principal component analysis (PCA) and Multivariate Variational Mode Decomposition (MVMD) for processing multichannel EEG signals. The EEG's intrinsic mode functions are disentangled using MVMD (IMFs). Finally, the electrooculography (EOG)-related data was obtained by summing the IMFs that link to EOAs. By using PCA to analyse the EOG signals, the key components linked to EOG was isolated and removed. Using the PCA and MVMD inverse transforms, we were able to recover the original, uncontaminated EEG signals. In this approach, measurements showed a CRC of 0.632, an MSE of 0.60, and an SNR of 57.8dB.