An Unsupervised Channel Selection Method for SSVEP-based Brain Computer Interfaces

Brain-computer interfaces (BCIs) provide an alternative communication channel for people with motor deficits that prevent normal communication. The underlying premise of a BCI is that a neuroimaging process such as electroencephalography (EEG) can be used to measure the user’s brain activity as signals. The obtained signals are analyzed to determine the user’s intended actions and a computer system can be used to replace voluntary muscle activity as a means of communication. The information transfer rate (ITR) of an algorithm used for determining the user’s intentions greatly affects the perceived practicality of the BCI system. Such algorithms are divided into two main categories, supervised and unsupervised. While the former achieves higher ITR, the latter is most useful when the user is unable to be involved in the calibration process of the BCI system. In our paper, we introduce an unsupervised algorithm for steady-state visual evoked potential (SSVEP)-based BCIs. Our algorithm works in three steps: (i) it selects multiple sets of electroencephalogram channels, then (ii) applies a feature extraction method to each one of these channel sets. As its final step, (iii) it combines the extracted features from these channel sets by performing a majority vote, yielding a classification. We evaluate the ITR attained using our proposed method on a dataset of 35 subjects using three different feature extraction methods. We then compare these results to existing methods in the literature that use a single channel set without a majority vote. The proposed method indicates an improvement for at least 7 subjects