A robust and subject-specific sequential forward search method for effective channel selection in brain computer...

Area of Science:

• Biomedical Engineering
• Neuroscience
• Signal Processing

Background:

• Electroencephalography (EEG) signals are crucial for brain-computer interfaces (BCI).
• Multi-channel EEG systems can lead to redundant information and increased computational complexity.
• Selecting only effective channels may enhance BCI performance and classification accuracy (CA).

Purpose of the Study:

• To develop a robust and subject-specific method for effective channel selection (ECS) in BCI.
• To improve classification accuracy while reducing the number of channels and computational complexity.

Main Methods:

• Proposed a robust and subject-specific sequential forward search method (RSS-SFSM) for ECS.
• ECS procedure sequentially searches channels to maximize CA on a validation set, repeated 100 times to identify effective channels.
• Applied RSS-SFSM to two datasets to demonstrate reliability and robustness.

Main Results:

• The proposed RSS-SFSM improved average CA by 15.98%.
• Reduced the number of channels and computational complexity by an average of 71.53%.
• Achieved superior results compared to existing methods in both CA improvement and channel reduction.

Conclusions:

• The RSS-SFSM demonstrates significant potential for identifying effective EEG channels for BCI applications.
• Features extracted using Hilbert transform and sum derivative methods were effectively classified by support vector machine.
• The method offers a promising approach for optimizing BCI system efficiency and accuracy.