A weighted EEG-fNIRS integration model enhances cortical representation of inter-muscular coordination

Inter-muscular coordination is closely linked to cortical activation and is essential for effective motor control. However, the relationship between cortical activity and inter-muscular coordination has not been thoroughly investigated, partly due to insufficient neural information. This study proposes a weighted EEG-fNIRS (Electroencephalography-functional Near-Infrared Spectroscopy) integration model to enhance the cortical representation of inter-muscular coordination. EEG and fNIRS data were collected from 15 participants performing one-dimensional (1D) and two-dimensional (2D) myoelectric-controlled interface (MCI)tracking tasks. These tasks were driven by myoelectric signals generated by the isometric contraction of specific muscles, including the biceps brachii and triceps brachii. The integration method involves extracting hybrid time-phase-frequency features from EEG signals, and it computes their classification accuracy to generate dynamic weights. These weights are then used to modulate fNIRS hemodynamic signals. To establish a baseline, representing a simplified reference model, where constant weights were calculated as the average of dynamic weights across time points. The classification accuracy of the time-phase-frequency features, serving as task-related weights, was higher than that of single features, achieving the highest within-class similarity (1DMCI: F = 5.08, p < 0.001; 2DMCI: F = 5.63, p < 0.001). Compared to the baseline model, the weighted integration model demonstrated higher within-class similarity (1D: p = 0.018, F = 8.38; 2D: p = 0.011, F = 10.46) and improved task discrimination by reducing irrelevant channels. These findings demonstrate that the weighted integration model effectively enhance the cortical representation of inter-muscular coordination, and has promising applications in brain research and clinical rehabilitation.