Spatial smoothing of canonical correlation analysis for steady state visual evoked potential based brain computer...

Area of Science:

• Neuroscience
• Biomedical Engineering
• Signal Processing

Background:

• Brain-computer interfaces (BCIs) enable communication via brain activity.
• Steady-state visual evoked potentials (SSVEPs) are a promising BCI paradigm using EEG.
• Standard Canonical Correlation Analysis (CCA) struggles with low classification accuracy for short EEG signals.

Purpose of the Study:

• To develop a novel regularization technique for CCA in SSVEP-based BCIs.
• To enhance the robustness and accuracy of CCA with limited signal data.
• To improve BCI performance under short signal length conditions.

Main Methods:

• Proposed a regularization method to create spatially smooth CCA spatial filters.
• Designed the spatial filter using a graph Fourier transform of 3D electrode coordinates.
• Compared the classification accuracy of the proposed regularized CCA against standard CCA.

Main Results:

• The proposed regularized CCA demonstrated superior performance compared to standard CCA.
• Significant improvements in classification accuracy were observed under short signal length conditions.
• Spatially smooth filters provided increased robustness for EEG signal processing.

Conclusions:

• The proposed regularization technique effectively addresses the limitations of standard CCA for short EEG signals.
• This advancement holds potential for more reliable and efficient SSVEP-based BCIs.
• Spatially smooth CCA filters are crucial for robust BCI performance with limited data.