The Graz Brain-Computer Interface

Abstract

Brain-computer interface (BCI) research at the Graz University of Technology started with the classification of event-related desynchronization (ERD) [36, 38] of single-trial electroencephalographic (EEG) data during actual (overt) and imagined (covert) hand movement [9, 18, 40]. At the beginning of our BCI research activities we had a cooperation with the Wadsworth Center in Albany, New York State, USA, with the common interest to control one-dimensional cursor movement on a monitor through mental activity [69]. With such a cursor control it is in principle possible to select letters of the alphabet, create words and sentences and realize a thought-based spelling system for patients in a complete or incomplete “locked-in” state [68]. At that time we already analyzed 64-channel EEG data from three patients who had accomplished a number of training sessions with the aim to search for optimal electrode positions and frequency components [38]. Using the distinction sensitive learning vector quantizer (DSLVQ) [54] it was found that for each subject there exist optimal electrode positions and frequency components for on-line EEG-based cursor control. This was confirmed recently by BCI studies in untrained subjects [2, 58].