Robust Decoding of Upper-Limb Movement Direction Under Cognitive Distraction With Invariant Patterns in Embedding...

Area of Science:

• Neuroscience
• Biomedical Engineering
• Machine Learning

Background:

• Motor brain-computer interfaces (BCIs) are crucial for rehabilitation and prosthetics.
• Decoding upper-limb movements using electroencephalography (EEG) is challenging due to cognitive distraction.
• Existing noninvasive BCIs often struggle with performance degradation under varying attentional states.

Purpose of the Study:

• To develop a robust EEG decoding model for upper-limb movement direction.
• To address the impact of cognitive distraction on BCI performance.
• To enhance the practical applicability of BCIs in real-world scenarios.

Main Methods:

• Proposed a novel decoding model utilizing invariant patterns within an embedding manifold.
• Reconstructed a low-dimensional manifold to characterize upper-limb movements from EEG signals.
• Transferred neural activity patterns from brain functional connectivity (FC) to a manifold subspace.

Main Results:

• The proposed decoding model demonstrated superior robustness on a mixed dataset encompassing both attentive and distracted states.
• Achieved higher decoding performance compared to baseline methods under cognitive distraction.
• Preserved essential movement-related information within the manifold subspace.

Conclusions:

• The novel decoding model offers improved robustness against cognitive distraction in EEG-based BCIs.
• The approach provides insights into a uniform mechanism underlying movement-related EEG signals.
• This research enhances the practicality of BCIs for real-world motor rehabilitation and control.