Continuous Bimanual Trajectory Decoding of Coordinated Movement From EEG Signals

Area of Science:

• Neuroscience
• Biomedical Engineering
• Machine Learning

Background:

• Bimanual coordination is crucial for many daily activities.
• Simultaneously decoding bimanual movement trajectories from electroencephalogram (EEG) signals remains a challenge.
• Existing brain-computer interface (BCI) paradigms often focus on unimodal control.

Purpose of the Study:

• To propose and validate a novel bimanual BCI paradigm for reconstructing continuous hand trajectories from EEG.
• To investigate the feasibility of simultaneously decoding bimanual movement position and velocity.
• To evaluate the performance of a deep learning model for bimanual trajectory decoding.

Main Methods:

• Developed a novel bimanual BCI paradigm involving reaching tasks to different targets.
• Collected EEG data during coordinated bimanual movements.
• Employed a multi-task deep learning model integrating EEGNet and Long Short-Term Memory (LSTM) network for decoding.
• Evaluated decoding performance using correlation coefficient (CC) and normalized root mean square error (NRMSE).

Main Results:

• The proposed model achieved significant decoding performance for both position (CC=0.54, NRMSE=0.22) and velocity (CC=0.42, NRMSE=0.23).
• Decoding accuracy was significantly superior to chance level (p<0.05).
• The model outperformed other commonly-used methods in continuous trajectory decoding.

Conclusions:

• Demonstrated the feasibility of simultaneously decoding continuous bimanual trajectories from EEG signals.
• Highlights the potential of this bimanual BCI for advanced human-computer interaction and assistive technologies.
• Suggests future research directions in enhancing decoding accuracy and exploring diverse coordinated movements.