Distance- and speed-informed kinematics decoding improves M/EEG based upper-limb movement decoder accuracy

Area of Science:

• Neuroscience
• Biomedical Engineering
• Signal Processing

Background:

• Brain-computer interfaces (BCIs) aim to restore function for individuals with paralysis.
• Decoding movement kinematics from brain activity is a key BCI research area.
• Noninvasive electroencephalography (EEG) and magnetoencephalography (MEG) studies show promise in accessing movement information.

Purpose of the Study:

• To investigate if combining directional and nondirectional kinematic information enhances BCI decoding accuracy.
• To assess the neural information associated with both types of kinematics for improved functional restoration.

Main Methods:

• Reanalyzed EEG and MEG data from 34 healthy participants performing executed and observed tracking movements.
• Decoded 2D movement trajectories using low-frequency M/EEG signals.
• Compared unscented Kalman filter (UKF) modeling nonlinear kinematics against linear Kalman (KF) and Wiener filters.

Main Results:

• Posterior-parietal, parieto-occipital, and sensorimotor areas encoded kinematic information.
• UKF achieved higher decoding accuracy (0.49 executed, 0.36 observed) compared to linear filters.
• UKF optimized signal-to-noise ratio and trajectory amplitude matching.

Conclusions:

• Directional and nondirectional kinematic information are simultaneously detectable in low-frequency M/EEG signals.
• Combining both kinematic types significantly improves decoding accuracy over linear methods.
• This finding advances BCI capabilities for functional restoration.