Abstracts

Rationale: In recent years, digital electroencephalography (EEG) analysis methods such as dipole analysis, wide-band EEG analysis, and qEEG (quantitative EEG) have been actively developed for secondary analyses of EEG data. However, the accuracy of these methods is largely dependent on the level of artifactual contamination of the EEG data. In order to improve this accuracy, we devised a novel method to remove contaminating artifacts from the EEG signals. Methods: The EEG was recorded  from human volunteers with a digital electroencephalograph (manufactured by Nihon Koden Co., Ltd.) , at a sampling rate of 500 Hz, via silver-contact scalp electrodes arranged on the International 10–20 method. The collected EEG data were decomposed and reconstructed with empirical mode decomposition (EMD) software developed in-house in MATLAB R2014a (MathWorks). We attempted to remove various contaminating artifacts produced by muscle activity(electromyogram), heart function(electrocardiogram), eye movements,  and sweating, among others. Results: The EMD software successfully scrubbed the artifactual contamination from the EEG data. Although the original EEG waveforms were clearly altered by bandpass filtering, EMD reconstruction had a significantly reduced effect on the source EEG waveforms, even taking time to analyze. Conclusions: Purification of EEG data using EMD is a potentially powerful support tool for secondary EEG analysis. However, it will be necessary to reduce the analysis time of the EMD, perhaps via utilizing the graphics processor, for which we plan to upgrade the program in the future. Funding: This abstract was supported by the Japan Science Foundation Science Research Fund Board C.