Research On The Portable Acquisition And Processing Of Single-Channel EEG

Brian is the most useful decision-making control system in human body with the most complex structure, the main performances of its advanced function are memory, emotion, perception, thinking and other natural phenomena. Electro-encephalogram (EEG) reflects a large amount of information about the physiological activities of the neurons and cells of brain, and it is an important carrier which the research about cognitive science, brain physiology, and clinical diagnosis of brain diseases are based on. Brain-computer interface (BCI) is an important part of brain research aimed at completing information exchange between brain and peripherals, and to realize the control of peripherals by measuring the potential of comprehensive scalp EEG, with a very broad application prospects.The core research of BCI is about EEG acquisition, analysis and processing. Due to the current EEG acquisition systems available in the market which are very expensive with a huge size, high power consumption, and complicated operation. All of these defects may limit the dissemination of the research about BCI systems. Based on NeuroSky’s TGAM (ThinkGear Asic Module) PCB module, this paper has developed a portable EEG data acquisition system with single-channel. A software which can be used to receive the EEG signal and display EEG wave real-time was also developed to save EEG dataset. The current BCI experiments are carried out generally based on EEG signal of C3 and C4 electrodes. Both of these electrodes request to be wet and conductive paste is also needed in addition, which will increase the burden of experiments. This paper has developed a portable single-channel EEG acquisition system, which can acquire forehead potential of human body easily using a dry electrode composed of Agcl object. This EEG acquisition system has many advantages, such as simple operation, stable performance, low cost and so on. It also makes reliable collection of EEG signal in the ubiquitous environment become possible.The EEG signal acquired using devices always contains lots of artifacts, such as electro-oculogram (EOG), which is caused by blinking of the subject unconsciously. The existing methods to remove artifacts from EEG usually require multi-channel EEG and known EOG signals as reference. These conditions are the bottlenecks of the application of portable single-channel EEG acquisition systems. In this paper, a method to remove artifacts from EEG adaptively based on Empirical Mode Decomposition and Independent Component Analysis (EMD-ICA) will be proposed, which is very suitable for the preprocessing of the single-channel portable EEG signal acquisition system. The proposed method will be used to deal with the EEG signal acquired using the designed single-channel EEG acquisition system factually in comparison with the classic wave-ICA (WICA) algorithm in time and frequency domain respectively. The experimental result shows that these two algorithms can remove the EOG artifacts from the single-channel EEG signal effectively, and the proposed method can obtain better denoising performance with retaining the real EEG information more effectively. This method can be used in the portable EEG acquisition for EEG de-noising pretreatment without EOG as reference signal.Based on the designed portable single-channel EEG acquisition system, this paper carries out an exploratory study on the application of the classification between the left and right hand movement imagination. With the reference to experimental paradigm used for collecting motor imagery EEG signal in the international BCI competition, an EEG signal acquisition experiment is designed to get movement imagination EEG of the left and right hand, and collecting a total of 400 groups (about 200 groups in each kind of imagination movement) of male and female motor imagery EEG signal. This paper also studies on EEG feature extraction and classification algorithm, HHT transform algorithm is used to extract time-frequency characteristic of movement imagination EEG signal, and the sample entropy is used as a measure of its complexity. Both of these features are used as characteristic parameters of the imaginations between left and right hand movement, with the support vector machine model used for identification and classification, and finally 86.5% of correct classification rate is achieved. The research results have certain theoretical reference value, and it is very helpful to promote the portability and popularization of brain-computer interface equipments.