A P300-Based BCI System Using Stereoelectroencephalography and Its Application in a Brain Mechanistic Study

Area of Science:

• Neuroscience
• Biomedical Engineering
• Signal Processing

Background:

• Stereoelectroencephalography (SEEG) offers high signal-to-noise ratio (SNR) intracranial recordings from both superficial and deep brain structures.
• SEEG's capabilities present a promising avenue for developing efficient brain-computer interfaces (BCIs) and advancing our understanding of brain activity.

Purpose of the Study:

• To implement and evaluate a P300-based brain-computer interface (BCI) utilizing SEEG signals.
• To investigate the performance of SEEG-based BCIs in spelling tasks.

Main Methods:

• A P300-based BCI was developed using SEEG signals elicited by a single-character oddball paradigm.
• Feature vectors from five SEEG contacts were processed using a Bayesian linear discriminant analysis (BLDA) classifier for target character prediction.
• Thirteen epilepsy patients with SEEG implants participated in the online spelling experiments.

Main Results:

• The SEEG-based BCI achieved an average online spelling accuracy of 93.85% among 13 participants.
• High performance was maintained even with single-channel SEEG signal analysis, indicating system robustness.
• Optimal decoding accuracy was observed from SEEG contacts located in the visual ventral pathway, including the fusiform gyrus (FG) and lingual gyrus (LG).

Conclusions:

• SEEG is a viable and effective modality for high-performance P300-based BCIs.
• The visual ventral pathway, particularly FG and LG, plays a crucial role in SEEG-based P300 BCI.
• These findings offer valuable insights for P300 mechanism research and BCI development.