HEGNet: EEG and EMG fusion decoding method in motor imagery and actual movement

The widespread adoption od brain-computer interface (BCI) has been hindered by the limited classification accuracy of electroencephalography (EEG) signals alone. This study proposes a novel BCI model, HEGNet, that addresses this challenge by fusing EEG and electromyography (EMG) signals. HEGNet incorporates an EMG feature extraction component to mitigate the inherent instability and low signal-to-noise ratio limitations of relying solely on EEG data. Additionally, HEGNet employs a feature fusion module to dynamically adjust the focus on EEG and EMG features, thereby enhancing its overall robustness. These findings suggest that EMG information can serve as a valuable supplement to EEG data.