FPGA implementation of EEG system-on-chip with automatic artifacts removal based on BSS-CCA method

Abstract

This paper presents an automatic muscle artifacts removal system for multi-channel electroencephalogram (EEG) applications. Since EEG signals are very weak and highly sensitive to the environment, they are easily contaminated by noises and artifacts. To get clean and usable EEG signals for brain-computer interface (BCI) applications, we should acquire these signals from the human brain without artifacts. Recently, Blind Source Separation (BSS) technique based on Canonical Correlation Analysis (CCA) was proposed to reconstruct clean EEG signals from recordings by removing muscle artifacts components. To enhance the feasibility and reliability of BCIs, EEG processing systems used for BCIs should be more portable and signals should be acquired in real-time without artifacts. To match with these requirements, a hardware design of the artifacts removal system is adopted for artifacts extraction. The performance of eye-blink and muscle artifacts elimination is evaluated through the correlation coefficients between processed and pure EEG signals. The experimental results show that the average correlation coefficients for eye-blink and muscle elimination are 0.9341 and 0.8927 respectively.