Abstract
Motor imagery-based brain-computer interfaces (MI-BCIs) represent a significant advancement in neurorehabilitation, yet their clinical efficacy is often limited by individual variability and an incomplete understanding of inter-regional brain communication. This study investigates the neural dynamics of three distinct motor imagery (MI) tasks: traditional static hand imagery (tMI), rotation movement imagery (rmMI), and bimanual coordination imagery (bcMI). Using Transfer Entropy (TE) analysis on EEG data from 14 healthy subjects, we demonstrate that bcMI significantly enhances directional information flow from parietal to prefrontal regions and strengthens contralateral motor-spatial connections. These results suggest that bimanual imagery paradigms can more effectively integrate motor-cortex functions, offering a superior framework for MI-BCI therapy.
Introduction
MI-BCIs facilitate rehabilitation by allowing patients to mentally rehearse movements, thereby triggering neuroplasticity. However, current protocols often rely on simple, one-handed tasks that may not fully engage the complex motor networks required for functional recovery. Two major hurdles persist: the inconsistency of brain signal patterns across individuals and a lack of clarity regarding how the motor cortex communicates with other brain regions during imagery. Given that physical bimanual training often yields better clinical outcomes than unilateral training, this research explores whether bimanual coordination imagery can induce more robust functional connectivity.
Methods
The study involved 14 healthy, right-handed participants (mean age 23.2). Brain activity was recorded via a 32-channel EEG system while subjects performed three MI tasks prompted by specific visual cues:
- Traditional MI (tMI): Imagining a static hand.
- Rotation Movement MI (rmMI): Imagining a single hand performing a forward rotation.
- Bimanual Coordination MI (bcMI): Imagining both hands operating a pedal exerciser.
Data analysis utilized Filter Bank Common Spatial Pattern (SB-CSP) and Linear Discriminant Analysis (LDA) for task classification. Functional connectivity was assessed using Transfer Entropy (TE), an information-theoretic measure capable of detecting the direction and dynamics of information flow between brain regions, identifying which areas "drive" others during the task.
Results
Classification and Neural Activation
All three imagery tasks achieved high classification accuracy: 92.1% for tMI, 91.5% for rmMI, and 87.1% for bcMI. This indicates that the increased complexity of bimanual feedback does not significantly impede the BCI's ability to recognize motor intent. Furthermore, all tasks successfully evoked Mu-rhythm (alpha) event-related desynchronization (ERD) in the motor cortex, confirming active engagement in motor imagery.
Enhanced Connectivity
The core finding of the study was the significant boost in functional connectivity during the bcMI task. TE analysis revealed a "parietal-prefrontal highway," with increased information flow from the parietal area (Pz) to bilateral prefrontal areas (F3, F4). This suggests that bimanual imagery more effectively recruits higher-level spatial processing and motor planning circuits. Additionally, bcMI promoted stronger contralateral connections, integrating the motor cortex with spatial coordination regions across alpha, beta, and gamma bands.
Discussion and Conclusion
The results indicate that bimanual coordination imagery increases spatial information processing and motor planning compared to traditional unilateral tasks. By promoting more effective connections across diverse brain regions, the proposed bcMI-BCI framework better integrates motor-cortex functions. This approach holds significant potential for enhancing MI-BCI rehabilitation protocols, particularly for stroke patients requiring complex motor recovery. Future work will focus on validating these findings in clinical populations to ensure these connectivity gains translate to improved motor recovery.
Summary updated