Neural time-series prediction preprocessing meets common spatial patterns in a brain-computer interface

This paper shows for the first time how a popular and successful filtering approach, known as the common spatial patterns (CSP) approach, compares to the neural time series prediction preprocessing (NTSPP) approach when applied in a 2-class EEG-based brain-computer interface (BCI), either using 2 or 60 EEG channels. Additionally, a novel NTSPP-CSP combination is developed to produce a 2-channel BCI system which significantly outperforms either approach operating independently and has the potential to outperform a 60 channel BCI involving the CSP approach with no NTSPP. The advantages of reducing the number of EEG channels being a reduction in the time used to mount electrodes and reducing the obtrusiveness of the electrode montage for the user. It is also shown that NTSPP can improve the potential of employing existing BCI methods with no subject-specific parameter tuning. Non subject-specific spectral filters are also employed with both approaches and tested with four different classifiers.