Title: Classifying brain-computer interface features based on statistics and density of power spectrum

Authors: Islam A. Fouad; Tareq Hadidi

Addresses: Biomedical Technology Department, Salman Bin Abdulaziz University, Al-Kharj, Saudi Arabia ' Biomedical Technology Department, Salman Bin Abdulaziz University, Al-Kharj, Saudi Arabia

Abstract: The use of Electroencephalography (EEG) signals as a vector of communication between men and machines represents one of the current challenges in signal theory research. The principal element of such a communication system, more known as 'brain-computer interface (BCI) ', is the interpretation of the EEG signals related to the characteristic parameters of brain electrical activity. In this paper, two feature extraction methods are applied: 'statistics method' and 'power spectral analysis (PSD)'. Then, two classification methods on a data set of BCI were compared: 'minimum distance classifier' and 'k-nearest neighbour classifier' to get the best results of discrimination between up and down movements. By applying the 'statistics method', it gives good results in both training data and test data. Also, the best classifier was the minimum distance for the training data and voting k-nearest neighbour for the test data.

Keywords: brain-computer interface; BCI; electroencephalography; EEG signals; feature extraction; classification; statistics; power spectral density; PSD; minimum distance classifier; k-nearest neighbour.

DOI: 10.1504/IJBET.2015.069849

International Journal of Biomedical Engineering and Technology, 2015 Vol.18 No.1, pp.1 - 13

Received: 25 Sep 2014
Accepted: 21 Dec 2014
Published online: 14 Jun 2015 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article