基于奇异谱熵的脑电意识任务识别方法的研究

奇异谱分析是脑电信号研究的一种新方法。脑电信号的奇异谱熵可以反映脑电的特征,它有助于研究大脑的动力学行为。时变脑电信号所对应的奇异谱熵时间序列能很好地反映出事件相关去同步(ERD)和事件相关同步(ERS)现象,因此可以提取人脑想象左右手运动任务时的特征,最终利用K-近邻模式分类方法对想象左右手运动任务进行有效的分类决策。最后对国际脑机接口竞赛2003相关数据进行了测试,最高准确率达到85.16%,最大互信息达到0.48。测试结果说明,基于奇异谱熵的脑电信号特征,可以作为脑电意识任务的有效分类依据。

On the Classification of Consciousness Tasks Based on the EEG Singular Spectrum Entropy

Singular Spectrum Analysis is a new method of studying the EEG signals. The singular spectrum entroy can reflect the features of EEG and contribute to the study of the dynamic behavior of EEG. The singular spectrum entroy features can effectively reflect the ERD/ERS time sequence changes, and be used as the feature parameters for the left and right hand motor imaginary tasks. Finally, a K-neighbors pattern recognition method is applied to optimizing the decision of discriminating the left and right hand motor imaginary tasks. The results of simulation on the data of the international brain computer interface contest in 2003 show that the maximum identification rate is 85.16% and the maximum mutual information is 0.48. The feature based on the singular spectrum entroy of EEG can make the classification of consciousness tasks more effective.