By Topic

Continuous EEG classification for a self-paced BCI

Sign In

Cookies must be enabled to login.After enabling cookies , please use refresh or reload or ctrl+f5 on the browser for the login options.

Formats Non-Member Member
$31 $13
Learn how you can qualify for the best price for this item!
Become an IEEE Member or Subscribe to
IEEE Xplore for exclusive pricing!
close button

puzzle piece

IEEE membership options for an individual and IEEE Xplore subscriptions for an organization offer the most affordable access to essential journal articles, conference papers, standards, eBooks, and eLearning courses.

Learn more about:

IEEE membership

IEEE Xplore subscriptions

3 Author(s)
Satti, A. ; Intell. Syst. Res. Centre, Univ. of Ulster, Derry, UK ; Coyle, D. ; Prasad, G.

Transferring electroencephalogram (EEG)-based brain-computer interface (BCI) systems from synchronous laboratory conditions to real-world applications and situations demands the continuous detection of brain patterns in which the user is in control of the timing and pace of the BCI instead of the computer. A self-paced BCI requires continuous analysis of the continuing brain activity, however, not only the intentional-control (IC) states have to be detected (e.g., motor imagery and imagination) but also the inactive periods, where the user is in a non-control state (NC). The nonstationary nature of the brain signals provides a rather unstable input resulting in uncertainty and complexity in the control. Intelligent processing algorithms adapted to the task at hand are a prerequisite for reliable self-paced BCI applications. This work presents a novel intelligent processing strategy for the realization of an effective self-paced BCI which has the capability to reduce noise as well as adaptation to continuous online biasing. A Savitzki-Golay filter has been applied to remove spikes/outliers while preserving the feature set structure. An anti-bias system is introduced which readjusts the classification output based on the brain's current and previous states. Furthermore, a multiple threshold algorithm is applied on the resultant unbiased classifier output for improved accuracy. These algorithms are tested on 4 real and 3 artificial datasets and results shown are considerably promising and demonstrate the significance of the proposed intelligent and adaptive algorithms.

Published in:

Neural Engineering, 2009. NER '09. 4th International IEEE/EMBS Conference on

Date of Conference:

April 29 2009-May 2 2009