Performance Analysis of Tri-channel Active Electrode EEG Device Designed for Classification of Motor Imagery Brainwaves for Brain ComputerInterface

摘要

This paper presents a modest, low cost method todesign a modular electroencephalography (EEG) signal acquisitionsystem for Brain Computer Interface (BCI) application.The system is based on active electrode EEG sensors along withArduino-Uno hardware, to acquire 3 channels for recording EEGsignals. In addition to that, a graphical user interface (GUI)was developed using Matrix Laboratory (MATLAB) for the purpose of signal processing and visualization. The performanceof designed EEG acquisition system was evaluated on the basisof classification of Motor Imagery (MI) brainwaves for imaginedleft-hand (LH) and right-hand (RH) movement. The designedsystem was used to record the MI brainwaves of 10 differentsubjects and the recorded data was split into a training set andtest set. Independent Component Analysis (ICA) was used for theremoval of EEG artifacts. The feature extraction was done usingWavelet Decomposition (WD) using Morlet as mother wavelet.The wavelet decomposed signal features like Maximum Power(MP) and Approximate Frequency (AF) was used as classificationfeatures for the classification of MI brainwaves. The classificationof MI brainwave signals was done using Linear DiscriminantAnalysis (LDA) which showed the accuracy of 81.6 percentagesfor testingdata set. Thus, the designed three channel active electrode EEGdevice could be used in various BCI applications.