皮质脑电图小波分析定位运动区的探索性研究

目的研究术中皮质脑电图(ECoG)小波分析对运动区皮质定位的可行性。方法利用小波变换,对ECoG信号进行多层分解和重构,提取4个主要频带(δ、θ、μ和β)重构信号的运动前后能量比(ERD)为特征量,并构造特定阈值进行分类,然后与相应手指弯曲运动数据进行对照比较,分析检测的正确率。结果d6子频带(μ频带)的运动前后ERD变化最明显;以40%为阈值进行分类,其定位运动区的正确率达到93%。结论通过小波分析对ECoG的特征进行提取和分类,可有效定位运动区皮质。

A exploration study of mapping human motor cortex by ECoG-based wavelet analysis

Objective To study the feasibility of mapping human motor cortex by cortical electroencephalogram （ECoG）-based wavelet analysis. Methods The ECoG signals were decomposed and reconstructed at multilevel by wavelet transform. Four main frequency strips （δ, θ, μ and β strip） were extracted from the reconstructed signals, and the energy ratio （ERD） of the signals before and after finger movement was considered as the feature parameter. A specific threshold was constructed for classification according to the ERD, and the correct rate was analyzed by comparing the finger flexion signals. Results d6 sub-band （μ band） had most obvious changes in ERD before and after finger flexion. The correct rate for detecting motor cortex was 93% according to the classification of 40% threshold of ERD in d6 sub-band. Conclusions ECoG feature extraction and classification can effectively determine the motor cortex by wavelet analysis.