Enhancing Classification Performance of Functional Near-Infrared Spectroscopy- Brain-Computer Interface Using...

Area of Science:

• Neuroscience
• Biomedical Engineering
• Signal Processing

Background:

• Brain-computer interfaces (BCIs) enable communication and control through brain signals.
• Functional near-infrared spectroscopy (fNIRS) offers a non-invasive method for monitoring brain activity.
• Accurate classification of fNIRS signals is crucial for effective BCI performance.

Purpose of the Study:

• To present a novel methodology for enhanced classification of fNIRS signals.
• To improve the performance of two-class BCIs using motor imagery (MI) and mental rotation (MR) tasks.
• To demonstrate the feasibility of a high-classification-performance fNIRS-BCI.

Main Methods:

• fNIRS signals for MI and MR tasks were acquired from the motor and prefrontal cortex, respectively.
• Signals were filtered to remove physiological noise and modeled using the general linear model with adaptive least squares estimation.
• Multiple feature combinations of estimated coefficients were used for classification with a support vector machine.

Main Results:

• Classification accuracies for MI versus rest ranged from 79.5% to 84.1%.
• Classification accuracies for MR versus rest ranged from 83.7% to 87.8%.
• The proposed methodology achieved significantly higher average classification accuracy (p < 0.05) compared to conventional hemodynamic response methods.

Conclusions:

• The novel methodology significantly enhances fNIRS signal classification for BCI applications.
• The findings demonstrate the potential for developing high-performance fNIRS-based BCIs.
• This approach offers a promising direction for advancing brain-computer interface technology.