Crossing time windows optimization based on mutual information for hybrid BCI.

Area of Science:

• Neuroscience
• Biomedical Engineering
• Signal Processing

Background:

• Hybrid Brain-Computer Interfaces (HBCI) combine EEG and fNIRS to improve performance.
• Current methods synchronize signals using fixed time windows, potentially mismatching distinct modal characteristics.
• This mismatch can negatively impact classification accuracy in tasks like mental arithmetic.

Purpose of the Study:

• To propose a novel crossing time windows (CTW) optimization for HBCI.
• To address the signal mismatch issue in hybrid BCIs.
• To enhance classification performance for mental arithmetic tasks.

Main Methods:

• EEG and fNIRS signals were segmented independently using sliding time windows.
• Crossing time windows (CTW) combined independently selected EEG and fNIRS segments.
• Features were extracted using Filter Bank Common Spatial Pattern (FBCSP) and statistical methods.
• Mutual information selected optimal windows based on feature discrimination.
• Fisher Lasso feature selection (FLFS) and Linear Discriminant Analysis (LDA) were used for classification.

Main Results:

• The proposed CTW method achieved a classification accuracy of 92.52 ± 5.38% for mental arithmetic tasks.
• This accuracy is significantly higher than other tested methods.
• The results demonstrate the effectiveness of optimizing time windows independently for each modality.

Conclusions:

• The novel crossing time windows optimization method enhances HBCI performance.
• Independent signal segmentation and optimization improve classification accuracy.
• This approach offers a more robust and accurate solution for mental arithmetic-based BCIs.