Robust EEG Channel Selection across Subjects for Brain-Computer Interfaces
- PDF / 1,305,361 Bytes
- 10 Pages / 600 x 792 pts Page_size
- 17 Downloads / 241 Views
Robust EEG Channel Selection across Subjects for Brain-Computer Interfaces 1 Thomas Navin Lal,2 Thilo Hinterberger,3 Martin Bogdan,1 ¨ Michael Schroder, 2 2 ¨ N. Jeremy Hill, Niels Birbaumer,3 Wolfgang Rosenstiel,1 and Bernhard Scholkopf 1 Department
of Computer Engineering, Eberhard-Karls University T¨ubingen, Sand 13, 72076 T¨ubingen, Germany Emails: [email protected], [email protected], [email protected]
2 Max
Planck Institute for Biological Cybernetics, Spemannstrasse 38, 72076 T¨ubingen, Germany Emails: [email protected], [email protected], [email protected]
3 Institute
of Medical Psychology and Behavioral Neurobiology, Eberhard-Karls University T¨ubingen, Gartenstrasse 29, 72074 T¨ubingen, Germany Emails: [email protected], [email protected]
Received 11 February 2004; Revised 22 September 2004 Most EEG-based brain-computer interface (BCI) paradigms come along with specific electrode positions, for example, for a visualbased BCI, electrode positions close to the primary visual cortex are used. For new BCI paradigms it is usually not known where task relevant activity can be measured from the scalp. For individual subjects, Lal et al. in 2004 showed that recording positions can be found without the use of prior knowledge about the paradigm used. However it remains unclear to what extent their method of recursive channel elimination (RCE) can be generalized across subjects. In this paper we transfer channel rankings from a group of subjects to a new subject. For motor imagery tasks the results are promising, although cross-subject channel selection does not quite achieve the performance of channel selection on data of single subjects. Although the RCE method was not provided with prior knowledge about the mental task, channels that are well known to be important (from a physiological point of view) were consistently selected whereas task-irrelevant channels were reliably disregarded. Keywords and phrases: brain-computer interface, channel selection, feature selection, recursive channel elimination, support vector machine, electroencephalography.
1.
INTRODUCTION
Brain-computer interface (BCI) systems are designed to distinguish two or more mental states during the performance of mental tasks (e.g., motor imagery tasks). Many BCI systems for humans try to classify those states on the basis of electroencephalographic (EEG) signals using machine learning algorithms. The input for classification methods is a set of training examples. In the case of BCI one example might consist of EEG data (possibly containing several channels) of one trial and a label marking the class of the trial. Classification methods pursue the objective to find structure in the data and as a result provide a mapping from EEG data to mental states. For some tasks the relevant EEG recording positions that lead to good classification results are known, especially when the tasks involve motor imagery (e.g., the imagination of limb movements)
Data Loading...
