EEG Source Imaging Enhances the Decoding of Complex Right-Hand Motor Imagery Tasks

Area of Science:

• Neuroscience
• Biomedical Engineering
• Signal Processing

Background:

• Sensorimotor-based brain-computer interfaces (BCIs) offer device control but often lack intuitive use due to a disconnect between imagined movements and output commands.
• Current BCI control signals may not reflect naturalistic actions, limiting user intent translation.
• High spatial resolution is needed to identify neural activity related to desired device actions.

Purpose of the Study:

• To enhance BCI performance by improving the decoding of complex motor imaginations.
• To develop and apply a novel technique for classifying natural hand/wrist manipulations.
• To address the cognitive disconnection in current BCI systems.

Main Methods:

• Extended existing EEG source imaging (ESI) techniques.
• Applied a novel classification method to decode four complex right-hand motor imaginations: flexion, extension, supination, and pronation.
• Compared ESI approach against traditional sensor-based methods.

Main Results:

• Achieved an 18.6% increase in individual task classification accuracy.
• Reported a 12.7% increase in overall classification accuracy using the ESI approach.
• Demonstrated superior performance of ESI over traditional sensor-based methods.

Conclusions:

• EEG source imaging (ESI) effectively enhances BCI performance for decoding complex motor imagery tasks.
• This research paves the way for more naturalistic and intuitive motor imaginations in BCIs.
• The findings support the broader adoption of noninvasive BCIs.