Global cortical activity predicts shape of hand during grasping

Area of Science:

• Neuroscience
• Biomedical Engineering
• Human-Computer Interaction

Background:

• Cortical field potentials are modulated by grasping kinematics.
• It remains unknown if scalp-recorded electroencephalography (EEG) can decode grasp kinematics.
• The optimal kinematic space (joint angles vs. synergies) for decoding is unclear.

Purpose of the Study:

• To determine if EEG can decode grasp kinematics from scalp activity.
• To compare decoding accuracy using joint angle velocities versus movement synergies.
• To investigate the spatial-temporal patterns of EEG activity during grasping.

Main Methods:

• Offline decoding of human EEG data during natural reach-to-grasp movements.
• Inferred hand joint angular velocities and synergistic trajectories.
• Measured decoding accuracy using correlation coefficient (r) between predicted and actual kinematics.

Main Results:

• Decoding accuracy for 15 hand joints was r = 0.49 ± 0.02.
• Decoding accuracies for the first three kinematic synergies were r = 0.59 ± 0.04, r = 0.47 ± 0.06, and r = 0.32 ± 0.05.
• EEG information about grasp type peaked at 250 ms post-movement onset, with early contralateral frontal-central and later central electrode recruitment.

Conclusions:

• Time-domain modulations in macro-scale EEG activity contain sufficient information to decode grasp kinematics.
• Decoding using kinematic synergies shows promise, particularly the first synergy.
• This non-invasive approach offers a potential pathway for developing brain-machine interfaces for prosthetic limb control.