Reconstruction and dynamic analysis of corticomuscular subnetworks reveal task-specific neural coordination in visuomotor control

Abstract

Background:

Understanding visuomotor coordination is critical for advancing behavioral neuroscience, particularly in elucidating fine hand control in daily activities. However, its dynamic organization in corticomuscular networks remains unclear.

New method:

This study proposes a novel framework for reconstructing and analyzing corticomuscular subnetworks from simultaneously recorded electroencephalography (EEG) and electromyography (EMG) signals during visuomotor tasks. Fifteen right-handed healthy subjects (9 males; mean age: $22.80\pm 1.01$ years) performed a grip-force tracking task under conditions with and without visual feedback. Dynamic corticomuscular networks were constructed with partial transfer entropy to capture evolving bidirectional information flow. Task-specific subnetworks were identified through non-negative matrix factorization and reconstructed based on activation-profile clustering with k-means. Subnetwork features were extracted, and correlation with behavioral performance was analyzed.

Results:

Eight activation profiles were identified without visual feedback, whereas three of these, involving more connections, were found with visual feedback. Moreover, subnetwork features involving efferent connections were correlated with tracking accuracy with visual feedback, while subnetworks involving afferent proprioceptive feedback were correlated with tracking accuracy without visual feedback.

Comparison with existing methods:

A subnetwork reconstruction technique was integrated with conventional network decomposition to deal with the over-segmentation problem, improving interpretability. To the best of our knowledge, this is the first study to decompose task-state dynamic cortico-muscular-cortical networks. Compared to full network analysis, the subnetworks showed stronger correlations with task performance.

Conclusion:

This study proposed a novel framework for analyzing dynamics of corticomuscular subnetworks and revealed distinct neural coordination strategies for visually guided and proprioception-driven motor control.