EEG Network Analysis System Based On MATLAB GUI

Brain science is incorporated into a new round of the five-year plan, indicating that China is about to set off a new wave of brain research. Based on Electroencephalogram(EEG) imaging technology and network analysis method is widely used to discover the brain’s secret, that is one of the main technology of brain science research. Also is an effective means of diagnosis and curing disease, such as neurodevelopmental disorders, psychotic disorders and neurodegenerative diseases. Meanwhile, it’s an important role for development of the computer, brain-computer interaction be used on intelligent robots, unmanned aerial vehicles, brain communications in the future. In short, brain science is not only to help improve human health, but also to promote the development of related industries, stimulate economic growth. Most studies have showed that the interactions among different brain areas contribute to the ability of brain in processing the corresponding information, while the diseases are also reported to be closely related to the connections among the whole brain. But now many EEG analysis software is still a lack of systematic brain network analysis function. Based on MATLAB platform, this paper achieves the graphical user interface system of brain network analysis, mainly including the following aspects of research:Firstly, over the past twenty-six years, the BCI team with technical accumulation in the analysis of EEG, the improvement and innovation of the algorithms is my thesis basis. Write technical documentation, build system framework, implement the core algorithms.Secondly, the system supports several mainstream data loading and display, according to the diversity of EEG data format. In EEG data, there are often bad guide, noise, artifact, etc., It is necessary to preprocess the load signal, to get the real clean EEG signals and ensure the correctness of the results of the subsequent processing. The system can remove and repair bad data, using REST method to refer data and convert any others reference to infinity. Use of Finite Impulse Response filter to extract the signal band and notch filter to remove 50 Hz power frequency interference. The artifact in the signal is separated by the improved ICA algorithm(Robust ICA) to separate the EEG signals, and remove artifact components.Thirdly, in order to study the characteristics of EEG from time domain, frequency domain and spatial distribution. The system uses Welch method to calculate the signal power spectrum estimation to reflect the energy distribution and frequency domain characteristics. Using Principal Component Analysis(PCA) and Independent Component Analysis(ICA) algorithm, to separate the main components and the independent components, then display the various components of the time domain waveform and spatial distribution. Using the Minimum Norm Solution(MNS) method obtain the EEG source localization results, and show corresponding activation area in the Italy 3-D real head model(cortical and scalp model).Finally, network analysis in the brain science research occupy a pivotal position. The thesis adopts correlation coefficient, coherence, phase synchronization and phase lag index methods to measure the undirected weighted network. As for directed network, using the Grainger Causality, Directional Transfer Function, Adaptive Birectional Transfer Function, Partial Directed Coherence to quantify. To determine the threshold and construct the neural network of the scalp after quantification the relationship between the signals. Respectively through the 2-D model and the 3-D model of the network topology reflects the connection strength and information interaction between nodes. Numerically, through the characteristic path length, clustering coefficient, global efficiency, local efficiency reflects the overall characteristics of the network. Using batching and statistical analysis function to eliminate the individual specificity of EEG and obtain statistical results.The system has been used to process P300 data, obtaining physiological basis results, and based on the results published the relevant papers. Therefore, it confirmed the analysis system with accuracy and practicality.