A novel paradigm based on radar-like scanning for directional recognition in event-related potentials based...

Area of Science:

• Neuroscience
• Biomedical Engineering
• Human-Computer Interaction

Background:

• Event-related potentials (ERPs) are crucial for brain-computer interface (BCI) directional control.
• Existing BCI paradigms struggle with scalability for numerous directional commands, requiring system reconfiguration.

Purpose of the Study:

• To introduce a novel radar-like scanning (RS) paradigm for high-resolution directional recognition in ERP-BCI systems.
• To overcome the limitations of existing paradigms in terms of scalability and command flexibility.

Main Methods:

• A radar-like scanning (RS) paradigm was developed, utilizing a sector-shaped visual stimulus to evoke ERPs for 32-directional recognition.
• An early-stopping strategy was implemented to enhance online experimental efficiency.
• EEGNet was used to analyze subject performance across different sector rotation periods (1s and 2s).

Main Results:

• The RS paradigm demonstrated strong negative ERP deflections in parietal, occipital, and temporoparietal regions.
• With a 2s rotation period and early stopping, the best subject achieved 87.50% accuracy and 1.64° mean absolute angle error.
• Subject-averaged accuracy reached 91.83% with a 11.25° error tolerance, with longer rotation periods improving performance.

Conclusions:

• The proposed radar-like scanning paradigm offers a scalable solution for fine-grained directional control in ERP-BCI systems.
• This novel approach is unaffected by the number of directional targets, demonstrating significant potential for BCI applications.