3D Input Convolutional Neural Network for SSVEP Classification in Design of Brain Computer Interface for Patient User

Area of Science:

• Neuroscience
• Computer Science
• Biomedical Engineering

Background:

• Brain-computer interfaces (BCIs) are crucial for restoring communication and control.
• Performance of BCIs relies on accurate signal classification and efficient information transfer.
• Steady-state visual evoked potentials (SSVEPs) are commonly used for BCI control.

Purpose of the Study:

• To evaluate the performance of 3-dimensional input convolutional neural networks (CNNs) for SSVEP classification.
• To compare 3D CNNs with 1D and 2D CNNs in a wireless EEG-based BCI system.
• To assess the impact of 3D CNNs on key BCI performance metrics.

Main Methods:

• Implemented 1D, 2D, and 3D CNNs for SSVEP signal classification.
• Conducted online experiments using a wireless EEG-based BCI system.
• Measured classification accuracy, information transfer rate, and user task completion time.

Main Results:

• The 3D CNN achieved an average classification accuracy of 93.75%.
• The 3D CNN reached an average information transfer rate of 58.35 bits/min.
• Both accuracy and information transfer rate significantly outperformed 1D and 2D CNNs.
• User task completion time was reduced using the 3D CNN.

Conclusions:

• 3D input CNNs represent a novel and state-of-the-art approach for SSVEP classification.
• The proposed 3D CNN method enhances BCI system performance.
• This advancement offers significant improvements for EEG-based brain-computer interfaces.