运动想象与其子过程相关脑电特征的数学分解和融合

Mathematical Decomposition and Fusion of EEG Features Related to Motor Imagery and Its Subprocesses

现有的基于运动想象的脑机接口(Motor Imagery-Brain-Computer Interface,MI-BCI)研究主要将每个运动想象过程作为一个整体来识别,但是这样的控制是不精细的,因此提出MI子过程的概念,通过把MI分解为子过程,对MI及其子过程相关脑电特征进行数学分解和融合,以提高解码的精细度、精度和鲁棒性。研究证明,一个连续完整的MI过程可以分解为若干子过程,反之,若干子过程可以融合为一个连续完整的MI过程。该研究可望为MI-EEG-BCI系统的精细控制提供思路。

The existing MI-BCI(Motor Imagery-Brain-Computer Interface) research based on motor imagery mainly identifies each motor imagery process as a whole, but such control is not refined. Therefore,this paper proposes the concept of MI sub-process, decomposes MI into sub-processes, and mathematically decomposes and fuses EEG features related to MI and its sub-processes to improve the refinement, accuracy and robustness of decoding. This study proves that a continuous and complete MI process can be decomposed into several sub-processes, and conversely, several sub-processes can be merged into a continuous and complete MI process. It is expected to provide ideas for the fine control of MI-EEG-BCI systems.