A Novel Real-time Algorithm Based on Phase-Locked Data Alignment for Continuously Controlled SSVEP-BCI

Brain-computer interfaces (BCIs) based on steady-state visual evoked potential (SSVEP) show good performance. However, algorithms always decode segments of electroencephalogram (EEG) and can only satisfy discrete output instructions, which limit the real-time continuous control of the BCI system. This article proposes a novel algorithm for SSVEP-BCI that can translate continuous EEG into control commands, achieving real-time monitoring of user intentions.

A phase synchronicity maximum strategy has been employed in this algorithm, which could capture a fixed-duration SSVEP epoch near any given moment, ensuring each trial is aligned with the phase of the potential corresponding template. Then, the algorithm utilized an update strategy of a small-step sliding window to recognize and output commands in approximately real-time.

We constructed an SSVEP-BCI system with continuous stimulation and recruited nine subjects. The results showed that the algorithm proposed in this study efficiently decoded continuously evoked SSVEP signals. The BCI's online average accuracy and ITR were 92.03% and 143.38 bits/min, respectively.

The proposed algorithm can decode SSVEP at any time theoretically, which improves command output density as well as maintains high recognition accuracy. This study provides novel methods for real-time control of external devices using SSVEP-BCIs and helps to develop BCIs that are more compatible with human control habits.