论文部分内容阅读
根据癫痫发作前后脑电(EEG)波形、能量、频率特征的不同,本文研究了两种小波分析和支持向量机(SVM)结合的脑电分类方法。一种直接利用EEG的波形特征对癫痫发作间歇期脑电和癫痫脑电进行分类,另一种采用EEG信号的波动指数和变化系数为特征进行分类;并比较了这两种方法分类的正确率。实验结果表明,两种方法均能有效区分间歇期脑电和癫痫脑电,以波动指数和变化系数为特征的方法具有更好的分类效果。
According to the characteristics of EEG waveform, energy and frequency before and after epileptic seizures, this paper studies two methods of EEG classification combining wavelet analysis and support vector machine (SVM). A direct use of EEG waveform characteristics of epileptic seizures intermittent EEG and epileptic EEG classification, the other using the EEG signal fluctuation index and coefficient of variation as a feature classification; and compared the accuracy of these two methods of classification . The experimental results show that both methods can effectively distinguish between intermittent EEG and epileptic EEG, and the methods characterized by fluctuating index and change coefficient have better classification results.