Phase Interval Value Analysis for the Motor Imagery Task in BCI

In neuroscience, phase is assumed to contain more important information about the neural activity than amplitude. However, the most exploited feature in electroencephalogram (EEG) based brain computer interface (BCI) is the amplitude change, phase has been largely ignored, and only phase locking values (PLV) has been introduced in EEG classification recently. In this paper, we define phase interval value (PIV) to explore the phase information of EEG from a new perspective and propose a computational model based on the ordered Parallel Factors (PARAFAC) algorithm to extract feature from multiway PIV data for single trial EEG classification. Application to the motor imagery task demonstrates that PIV is quite effective for EEG classification, providing significant and discriminative features in spatial and spectral dimension. PIV might become an important new tool in the analysis of EEG phase characteristic, and has the great potential use in BCI.