A Calibration-Free Hybrid Approach Combining SSVEP and EOG for Continuous Control

Area of Science:

• Neuroscience
• Biomedical Engineering
• Human-Computer Interaction

Background:

• Steady-state visually evoked potential-based brain-computer interfaces (SSVEP-BCI) are widely used for discrete control.
• Continuous SSVEP-BCI offers real-time command delivery but faces challenges with the transition state problem during gaze shifts.

Purpose of the Study:

• To develop a calibration-free Bayesian approach for continuous SSVEP-BCI by hybridizing SSVEP and electrooculography (EOG).
• To address the transition state problem and improve accuracy and speed in continuous BCI control.

Main Methods:

• Hybridized SSVEP and EOG signals for continuous BCI control.
• Utilized Canonical Correlation Analysis (CCA) for SSVEP detection and an adaptive threshold method for EOG-based saccade detection.
• Employed a Bayesian optimization approach to recognize new targets by integrating SSVEP and saccade data.

Main Results:

• Offline experiments showed superior continuous accuracy and reduced gaze-shifting time compared to existing methods (FBCCA, CCA, MEC, PSDA).
• Online experiments demonstrated significantly higher continuous accuracy (77.61% vs. 68.86%) and faster gaze-shifting time (0.93s vs. 1.94s) compared to CCA-based SSVEP-BCI.
• Participants reported a significant improvement in user experience with the proposed hybrid BCI.

Conclusions:

• The proposed hybrid Bayesian approach effectively enables calibration-free continuous BCI control.
• This framework enhances SSVEP and EOG integration, promoting plug-and-play BCIs for continuous applications.
• The study validates a novel method for improving the robustness and usability of SSVEP-BCI systems.