基于多分类SVM算法的癫痫发作自动检测方法

Automatic detection of seizure based on multiclass SVM algorithm

高性能的癫痫脑电信号自动检测方法对减轻医生负担并提高癫痫的诊断效率具有重要临床研究意义。论文提出了一种能够区分正常、癫痫发作和发作间期脑电信号的高性能三分类系统。采用Daubechies 4小波构成的4级提升式小波变换将脑电信号分解为不同子带信号,求得不同子带信号的近似熵、Teager能量、局部波动率、自回归系数、Hurst指数特征值;利用Fisher得分法进行特征选择,提高分类精度同时减小计算复杂度;基于二叉树多分类支持矢量机(support vector machine,SVM)对脑电信号分类,实现正常、癫痫发作和发作间期信号的自动检测。实验表明,系统的准确率、灵敏度和特异性均达到100%,优于现有的分类识别方法,提出的三分类系统具有良好的分类性能,为癫痫及癫痫发作的临床检测提供了较好参考价值。

An automatic detection of epileptic electroencephalogram(EEG)with high-performance is of clinical research significance in both relieving heavy workload of doctors and improving the diagnosis efficiency for epilepsy.A three-class classification system with high-performance for distinguishing normal,ictal,and interictal EEG signals is presented in this paper.Four-level lifting discrete wavelet transform using Daubechies order 4 wavelet is introduced to decompose the EEG signals into different sub-bands,and features with approximate entropy,teager energy,local volatility,autocorrelation coefficient,and hurst exponent are extracted.In order to reduce the computational complexity and increase the classification accuracy,fisher score is used for feature selection.Binary tree multiclass support vector machine(SVM)is used to classify the EEG signals into normal,ictal,and interictal for automatic detection.The performance of the designed three-class classification system is tested with publicly available epilepsy dataset.The results show that the three-class system achieves 100%accuracy,sensitivity,and specificity,respectively,which is superior to the state-of-the-art detection methods.With excellent classification performance,this three-class classification system contributes to the clinical detection of epilepsy and seizure.