Abstract:

BACKGROUND: The electroencephalogram signal can reflect the different physiological and pathological activity of brain, many noises are interfused into electroencephalogram signals during the collecting and analyzing process, such as eye movements, eye blinks, heart beats and muscle activities, which affects people’s right to analysis and process the electroencephalogram signal.
OBJECTIVE: To introduce an independent component analysis approach based on the extended information maximization in order to perform the noise removal in electroencephalogram signal.
METHODS: The iteration of extended information maximization was performed to obtain the separation matrix, and the independent component after noise removal was used to reconstruct the electroencephalogram signal that need to be recorded. The electroencephalogram signal after noise removal with Matlab simulation was observed, and the correlation between electroencephalogram signal and electrooculogram signal was compared.
RESULTS AND CONCLUSION: The independent component analysis approach based on extended information maximization could successfully remove the electrooculogram signal intervention from the electroencephalogram signal. The comparison of the power spectrum of the electroencephalogram signals before and after noise removal showed the independent component analysis approach based on extended information maximization could remove the frequency interference from the multi-channel electroencephalogram signals effectively and have no damage to other signals.

Key words: bone and joint implants, basic experiment of bone injury, electroencephalogram signal, noise, interference, noise removal, removal, information maximization algorithm, expansion, independent component analysis, neural network, brain disease