Sequential selection of window length for improved SSVEP-based BCI classification.

Area of Science:

• Neuroscience
• Biomedical Engineering
• Signal Processing

Background:

• Brain-computer interfaces (BCI) using steady-state visually evoked potentials (SSVEP) offer potential for assistive technologies and advanced control systems.
• Current SSVEP-BCIs face a speed-accuracy trade-off in classifying user attention to specific modulation frequencies.
• Fixed data window lengths may not optimally capture the dynamic nature of SSVEP signals.

Purpose of the Study:

• To develop and evaluate a novel sequential analysis strategy for adaptive window-length selection in SSVEP-BCI classification.
• To overcome the limitations of fixed-length data windows in optimizing the speed-accuracy trade-off for SSVEP-BCIs.

Main Methods:

• A variable-length data window strategy based on sequential analysis was developed.
• The adaptive strategy collects data until a confident classification decision can be made.
• Performance was compared against a fixed window-length method using electroencephalography (EEG) data from three participants viewing five distinct frequencies.

Main Results:

• The proposed variable-length window approach demonstrated superior performance compared to fixed-length methods.
• Canonical correlation analysis was employed as the classification algorithm.
• An average improvement of 43% in the classifier's information transfer rate was observed with the variable-length scheme.

Conclusions:

• Adaptive window-length selection using sequential analysis is an effective strategy for enhancing SSVEP-BCI performance.
• This method offers a significant improvement in the information transfer rate, addressing the speed-accuracy trade-off.
• The findings suggest a promising direction for developing more efficient and responsive BCIs.