Enhancement of Motor Imagery Brain Computer Interface Performance Using Channel Reduction Method based on Statistical Parameters

1st International Conference on Mechanical Electronic and Biosystem Engineering

Zanar Azalan, Mohd Shuhanaz^1 ; Paulraj, M.P.^2 ; Hamid Adom, Abdul^1

School of Mechatronic Engineering, Universiti Malaysia Perlis, Perlis, Arau, Malaysia^1

Sri Ramakrishna Institute of Technology, Tamilnadu, Coimbatore, India^2

关键词: Channel reduction;    Classification accuracy;    Classification performance;    Computational time;    Neural network (nn);    Recorded signals;    Standard deviation;    Statistical parameters;   

Others  :  https://iopscience.iop.org/article/10.1088/1757-899X/557/1/012016/pdf
DOI  :  10.1088/1757-899X/557/1/012016

来源: IOP

【 摘 要 】

In this paper, a novel method to reduce the number of EEG channels for a Motor Imagery-based Brain Computer Interfaced (BCI) system without compromising its performance is proposed. By reducing the number of EEG channels, the number of features can be reduced and this has to be achieved without sacrificing the classification accuracy and computational time of the BCI. EEG signals were recorded from 10 subjects using a 19-channel EEG amplifier. Higuchi Fractal features were extracted from the recorded signals and modelled using Neural Networks (NN). A simple statistical analysis based on standard deviation was then used for the channel reduction process. The classification accuracy of the NN model formulated with the 19 channels features were compared to that of the model with features selected using statistical method. From the results it was observed that using this approach, the number of EEG channels can be reduced up to 30% without sacrificing its classification performance.