Intelligent Feature Classification for a Practical Brain-Computer Interface
Background
Electroencephalogram (EEG) and other electrophysiological measures that reflect brain function can support non-muscular pathways for communication and control, commonly called brain-computer interfaces (BCIs). These are especially useful for those who are paralysed or have severe motor disabilities, such as people suffering from motor neurone diseases (MNDs) and spinal cord injury (SCI) [5][6]. A conservative estimate is that 1 in 3500 of the world population may suffer from a neuro-muscular disorder [7]. Beyond medical applications, a practical BCI offers an additional and independent communication channel to healthy users using brain activities alone, which has a range of promising applications such as computer games with intuitive control strategies [8] and advanced virtual reality (VR) scenarios [4]. Due to severe non-linearities and non-stationarities in brainwaves characteristics obtained from EEG signal, the current EEG-based BCI systems suffer from limited accuracy, are insufficiently robust for regular and sustainable use and are too expensive to maintain. At ISRC, we have developed promising algorithms for EEG signal pre-processing, feature extraction, and feature classification providing significant improvement in BCI performance [1][2][3]. Building on these promising works, this project aims to investigate how non-stationarities in EEG signal characteristics can be effectively accounted for in an automated way, so that a more practical BCI system could be developed.
Research Program
To accomplish the above aim, this project proposes to investigate novel intelligent techniques that can effectively account for non-linearities and non-stationarities in brainwaves characteristics making enhanced utilisation of EEG data obtained from subject training on motor-imagery related cognitive tasks. In particular, it will involve devising a novel feature classification scheme that consists of multiple robust classifiers and a classifier selection mechanism, so that in the on-line operation, the most appropriate classifier is automatically selected for feature classification at all time points. Multiple classifiers will be identified off-line for variable situations occurring due to physiological and other factors. The project will also involve a thorough evaluation of the efficacy of the newly developed BCI in the state-of-the-art ISRC BCI lab, when used by both disabled and able-bodied subjects to operate a specially designed asynchronously controlled fast virtual keyboard and a robotic device.
Anticipated outcomes
It is anticipated that a successful completion of the project will result into a practical and reliable BCI that can be used on a regular basis both by the disabled and healthy users.
References
[1]. Herman, P, Prasad, G, McGinnity, T.M. Coyle, D.H., “Comparative analysis of spectral approaches to feature extraction for EEG-based motor imagery classification”, IEEE Trans. on Neural systems and Rehabilitation Engineering, Vol. 16, No. 4, pp 317 – 326, Aug 2008.
[2]. Coyle, DH, Prasad, G, McGinnity TM, “A Time-Series Prediction Approach to Extracting Features for a Brain-Computer Interface”, IEEE Trans. On Neural systems and rehabilitation engineering, 2005, vol. 13, no. 4, pp. 461-467.
[3]. Coyle, DH, Prasad, G, McGinnity TM, “A time-frequency approach to feature extraction for a brain-computer interface with a comparative analysis of performance measures", EURASIP Journal of Applied Signal Processing, 2005, Vol. 2005, Issue 19, pp 3141-3151.
[4]. Pfurtscheller G, “Importance of Motor Imagery and of Feedback Observation of a Moving Object in BCI Research”, 2nd International BCI Workshop & Training Course, Biomed. Tech. 49 (2004), Erg. 1, pp 23-28.
[5]. Kuebler A., Kotchoubey, B, Kaiser, J, Wolpaw, JR, Birbaumer, N, “ Brain-computer communication : unlocking the locked in”, Psychology Bulletin, 2001, Vol. 127, pp 358-375.
[6]. Wolpaw JR, Birbaumer, N, McFarland DJ, Pfurtscheller, G, Vaughan, TM, “Brain-computer interfaces for communication and control “, Clinical Neurophysiology, 2002, Vol. 113, pp 767-791.
[7]. Emery, AEH, “Population frequencies of inherited neuro-muscular diseases – A world survey”, Neuro-muscular disorders, 1991, Vol. 1, No. 1, pp 19-29.
[8]. Krausz, G, Scherer, R, Korisek, G, Pfurtscheller, G, “Critical Decision-Speed and Information Transfer in the Graz Brain-Computer Interface”, Applied Psychophysiology and Biofeedback, 2003, Vol. 28, No.3, pp 233-240.
Personnel Involved
First Supervisor: Prasad, G Dr
Second Supervisor: Li, Y Dr
Collaboration: This project does not involve collaboration with another establishment