Design of a mental task-based brain-computer interface with a zero false activation rate using very few EEG electrode channels

To design a practical brain-computer interface, the high rate of false activation and the high number of necessary electrodes are two major problems that must be addressed. The objective of this study is to design a brain interface system that requires very few channels, has a zero false activation rate and a high true activation rate. To attain this objective, a brain-computer interface that is EEG-based and that is activated by mental tasks is proposed. The system is custom designed for each subject. For each subject, the most discriminatory mental task that yields a zero false activation rate is determined. By keeping the false positive rate at zero, the number of channels needed is reduced. We show that we can obtain a false positive rate of zero value and a true positive rate in the range of 71.96% to 77.61% with only three electrode channels. The dataset used was not collected in a self-paced paradigm; however, it is employed to show that the design of a self-paced interface is feasible. EEG signals of four subjects performing five mental tasks are used as data. Applying fast and simple approaches like the autoregressive modeling and the quadratic discriminant analysis as the feature extraction and classification methods, respectively, is another advantage of the present work.