DEVELOPMENT OF WEARABLE ELECTROENCEPHALOGRAM (EEG) HEADSET SYSTEM FOR ELIMINATING ARTIFACTS

• Main Author: Karomatul Masyhuroh, Fauza
• Format: Final Project
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/32351
• Institution: Institut Teknologi Bandung

Noninvasive EEG that commonly utilized for clinical has been developed to nonclinical applications, e. g. drowsiness detection, games, emotion recognition, and so forth. Regularly, subjects must recline and lessen their movement in order to avoid large artifact due to EEG electrode shift. However, that rule would be hardly applied to non-clinical EEG due to subject movement that could increase the artifacts existence. Artifacts denote any electrical signal generated by sources other than human brain, but captured by EEG data acquisition. To achieve those new purposes, EEG system has to be transformed from wired, stationary, and cumbersome technology that mostly used for clinical practice, to wearable and wireless system. EEG headset as a solution for those needs has been proposed. Nevertheless, many research groups revealed that its recorded EEG signal has lower quality and the system tends to be closed. Furthermore, the EEG headset offered gives fixed electrode position, so as it could increase the artifacts due to mismatch between head size and headset dimension. This research built a wearable, adjustable, and opened EEG headset prototype that provides less artifact signal. Moreover, a specific material which is abundant in Indonesia, strong, flexible, and eco-friendly was used to ease headset manufacture, in addition to fulfill environmental issues. Headset testing was processed by conducting an EEG experiment using both headset and without headset. EEG signal in relax condition was measured to get baseline signal. Afterwards, the subject of this experiment was asked to shake their head as an EEG signal that contains motion artifact. Both signal were compared using either qualitatively (based on signal pattern in time and frequency domain) or quantitatively (using correlation method). Further, to obtain much better EEG signal, Independent Component Analysis (ICA) was applied within signal processing for removing eye blink artifact that could not be handled by headset. As the result, qualitatively EEG signal using headset shows higher resemblance to the baseline pattern than recorded signal without headset for both time and frequency domain. Correlation method also presented the successful of EEG headset. Quantitatively, the coefficient correlation percentage of its EEG signal is higher than EEG signal without headset. Further, ICA method implementation gives a result that ICA effectively eliminate eye blink artifact.