Reconstructing three-dimensional hand movements from noninvasive electroencephalographic signals.

Area of Science:

• Neuroscience
• Biomedical Engineering
• Motor Control

Background:

• Noninvasive electroencephalography (EEG) is often considered insufficient for detailed neural decoding of complex movements.
• Previous studies primarily relied on invasive intracranial recordings for kinematic decoding.

Purpose of the Study:

• To challenge the assumption of EEG's limitations by decoding 3D hand velocity from scalp EEG during natural reaching.
• To investigate the potential of EEG for noninvasive brain-computer interfaces (BCIs) for motor prosthetics.

Main Methods:

• Continuous decoding of 3D hand velocity from 55-channel EEG data during a self-initiated, self-selected 3D center-out reaching task.
• Controlled eye movements to prevent confounding results.
• Analysis of sensor contributions and source localization using standardized low-resolution brain electromagnetic tomography (sLORETA).

Main Results:

• Achieved reasonable decoding accuracy of 3D hand velocity from scalp EEG, comparable to some intracranial studies.
• Identified significant contributions from sensors over the contralateral sensorimotor cortex.
• Localized current density sources to the contralateral precentral gyrus, postcentral gyrus, and inferior parietal lobule.
• Found a negative correlation between movement variability and decoding accuracy.

Conclusions:

• Scalp EEG can continuously decode 3D hand velocity during natural upper limb movements, challenging previous assumptions.
• The findings support the development of noninvasive neuromotor prostheses.
• Movement variability is a critical factor influencing BCI performance.