Novel hybrid brain-computer interface system based on motor imagery and P300

Area of Science:

• Neuroscience
• Biomedical Engineering
• Rehabilitation Technology

Background:

• Motor imagery (MI) is crucial for brain-computer interfaces (BCIs) in post-stroke rehabilitation.
• Early-stage MI training suffers from unreliable feedback due to insufficient data discriminability, potentially hindering patient motivation and progress.

Purpose of the Study:

• To propose and evaluate a novel hybrid BCI paradigm integrating MI and P300 detection.
• To enhance the performance and efficiency of BCI-based rehabilitation training in the critical early stages.

Main Methods:

• A hybrid BCI paradigm was developed, combining imagined Chinese character writing (MI) with P300 evoked by character strokes.
• A fusion method was implemented to correct unreliable P300 classifications using robust MI classifications.
• The system was tested on twelve healthy participants performing imagined writing tasks.

Main Results:

• The hybrid BCI paradigm significantly outperformed single-modality BCI systems.
• The fusion method achieved higher recognition accuracy compared to P300 alone (p<0.05) and MI alone (p<0.01).
• The proposed fusion approach demonstrated a reduction in the required training data size.

Conclusions:

• The hybrid MI-P300 BCI paradigm offers a promising solution for improving early-stage BCI training in stroke rehabilitation.
• Fusion of MI and P300 signals enhances classification accuracy and data efficiency, addressing key limitations of current MI-BCIs.