Multimodal 2D Brain Computer Interface.

Area of Science:

• Neuroscience
• Biomedical Engineering
• Signal Processing

Background:

• Brain-computer interfaces (BCI) enable control through neural signals.
• Non-invasive techniques like NIRS and EEG offer potential for BCI applications.
• Integrating multimodal brain recordings may enhance BCI performance.

Purpose of the Study:

• To evaluate the efficacy of multimodal brain signal recording for BCI control.
• To compare classification accuracy using individual versus combined NIRS and EEG features.
• To investigate motor imagery classification using NIRS, EEG, and tEEG.

Main Methods:

• Utilized Near Infrared Spectroscopy (NIRS), electroencephalography (EEG), and tripolar concentric ring electrodes (tEEG) for brain signal acquisition.
• Extracted features using signal slope (SS) from NIRS and power spectrum density (PSD) from EEG/tEEG (8-30Hz).
• Employed Linear Discriminant Analysis (LDA) for classifying motor imagery tasks (left hand, right hand, both hands, both feet).

Main Results:

• The highest classification accuracy reached 85.2% when integrating features from all three recording systems (NIRS, EEG, tEEG).
• Multimodal signal feature integration yielded a statistically significant improvement in classification accuracy (p = 0.0033) compared to EEG-only features.
• Individual feature sets showed varying classification performance, with combined features demonstrating superior results.

Conclusions:

• Multimodal brain signal recording significantly enhances BCI performance for motor imagery tasks.
• Combining NIRS and EEG data provides a more robust and accurate method for BCI control.
• Future research should explore advanced feature extraction and classification algorithms for multimodal BCI systems.