基于匹配跟踪的脑电睡眠纺锤波的时频分析

Using Matching Pursuit Algorithm to Analyze Electroencephalogram (EEG) Sleep Spindles in Time-Frequency Domain

匹配跟踪(MP)算法是一种冗余算法,通过信号x在所选择的最佳向量(最佳基)上的反复正交投影而逼近原信号.它不仅可以获得较好的时频分辨率,而且可以参数化描述任何类型的数据.在临床应用实践中已发现睡眠纺锤波与多种病症有关,因此,有效地检测出睡眠纺锤波具有重要的临床应用价值.文中根据脑电信号(EEG)睡眠纺锤波的特征,利用基于Gabor函数的MP算法对EEG中的睡眠纺锤波进行检测和分析,并给出了分析结果的时频表示.

Matching pursuit (MP) algorithm is a redundant algorithm, which approximates raw signals by an orthogonal projection on the selected best vectors (the best atoms) of signal x. It can get not only high time-frequency resolution, but also parametric description of all kinds of data structures. In clinical application practice, the EEG sleep spindles have been found to be relative to the multifold pathology, so it is of an important clinical application value to detect EEG sleep spindles effectively. In this paper, according to the feature of EEG sleep spindles, the matching pursuit (MP) algorithm based on Gabor function (the basis function) was used to detect and analyse EEG sleep spindles, and the time-frequency representation of the analysis results was given.