Decoding hand movement velocity from electroencephalogram signals during a drawing task.

Area of Science:

• Neuroscience
• Biomedical Engineering
• Rehabilitation Technology

Background:

• Decoding neural activity for limb movement is crucial for motor prosthesis control.
• Current methods often rely on invasive techniques, limiting widespread application.
• Non-invasive approaches like electroencephalogram (EEG) show potential but require further investigation for complex movements.

Purpose of the Study:

• To investigate the feasibility of decoding hand movement velocity using EEG signals.
• To explore decoding performance during a self-routed drawing task, a more naturalistic movement.
• To identify key EEG features and brain regions involved in hand velocity decoding.

Main Methods:

• Collected whole-scalp EEG data from five subjects during a sequential 4-directional drawing task.
• Applied spatial filtering to extract amplitude and power features across multiple frequency bands.
• Utilized Kalman filtering and smoothing algorithms to reconstruct hand movement velocity.

Main Results:

• Achieved average Pearson correlation coefficients of 0.37 (horizontal) and 0.24 (vertical) between measured and decoded velocities.
• Identified motor, posterior parietal, and occipital brain areas as most relevant for decoding hand velocity.
• Found that both slow potentials (0.1-4 Hz) and specific oscillatory rhythms (24-28 Hz) contain hand velocity information.

Conclusions:

• Demonstrated that hand movement velocity can be decoded from EEG signals during a drawing task.
• Provides evidence supporting the use of EEG and advanced decoding methods for neural control of motor prostheses.
• Highlights the potential of non-invasive neuroimaging for developing next-generation assistive devices.