Decoding Imagined 3D Arm Movement Trajectories From EEG to Control Two Virtual Arms-A Pilot Study.

Area of Science:

• Neuroscience
• Biomedical Engineering
• Human-Computer Interaction

Background:

• Current brain-computer interfaces (BCIs) for prosthetic/virtual arm control often rely on classifying distinct sensorimotor states.
• Decoding imagined 3D arm movement trajectories for naturalistic control is less explored, with existing studies limited to offline applications.

Purpose of the Study:

• To achieve real-time, online control of two virtual arms in 3D space using decoded imagined arm movement trajectories.
• To compare the efficacy of decoding imagined 3D arm movement trajectories against a filter-bank common spatial patterns (FBCSP) classification method.

Main Methods:

• Decoding 3D imagined arm movement trajectories from electroencephalography (EEG) power spectral densities (mu, beta, gamma oscillations) using multiple linear regression.
• Online control of two virtual arms towards three targets, with offline calibration and two online feedback blocks.
• Comparison with FBCSP-based multi-class classification (using mutual information selection and linear discriminant analysis).

Main Results:

• Offline decoding of imagined 3D arm movement trajectories achieved significantly above-chance accuracy (45%) for two subjects.
• Real-time control accuracy using decoded trajectories was at chance level (33.3%), but false-positive feedback improved performance for some subjects.
• FBCSP successfully distinguished left/right arm imagination (70%) but failed for three-target classification (33.3%).

Conclusions:

• Online control of virtual arms via decoded 3D imagined arm movement trajectories remains challenging, with current methods performing at chance level.
• Closed-loop feedback, even with false positives, may enhance real-time BCI performance.
• Identified sub-optimal aspects of the experimental paradigm and proposed improvements for future research.