基于共同空间模型的癫痫脑电检测预测的优劣

DETECTION AND PREDICTION OF EPILEPTIC EEG BY COMMON SPATIAL PATTERNS

从辅助临床诊断为出发点,对癫痫脑电波的数据进行分析,用统计模式识别的方法——共同空间模型,结合统计学习理论——支持向量机,对脑电波进行检测和分类,结果表明共同空间模型方法可以很好的区分正常和异常的脑电波.而且对于一些癫痫发作模式,能够找到发作之前的征兆脑电波,进行预测.最后,对该检测预测系统之优劣进行了讨论.

In epilepsy brain neurons produce abnormal electrical signals,which could be used for diagnosis.Electroencephalographs（EEG） were analyzed on both time and frequency domains.Statistic pattern recognition,common spatial patterns（CSP） and support vector machine（SVM） were used to classify EEG signals.It was suggested that special firing patterns may exist before epilepstic attacks.