Investigating ICA for EEG Electrode Optimization for The Differentiation Between Right-Hand and Left-Hand Movements

摘要

A bionic hand that is controlled by an electroencephalograph (EEG)-based brain computer interface (BCI) can aid motor impaired individuals to perform daily tasks. High-density EEG (128 electrodes) are suggested for the spatial resolution required to control these activities. This makes the system expensive, time-consuming to set up and uncomfortable for the user. This research explores the development of a novel electrode reduction method that combines independent component analysis (ICA) and features related to event-related desynchronization and synchronization (ERD/ERS) modulations to produce an optimised and reduced EEG electrode set. This method was tested for the differentiation between right-hand and left-hand movements. The results suggest that the optimal channel configuration produced was a 16-electrode configuration. The 16-electrode configuration obtained a classification accuracy of 70.51 %, using a linear support vector machine, which is a 12.01% loss in classification accuracy when compared to using the full 128-electrode set. This suggests that ICA could be used as a primary technique to reduce the number of electrodes of an EEG-based BCI controlling a bionic hand. The research also suggests that motor control information could be captured from widely distributed electrodes.