Abstract: Identifying seizures from EEG signals is a crucial tool in clinical diagnosis of epilepsy. However, the accuracy of manually labeling EEG signals is barely satisfactory. In this paper, a method based on functional brain network and TSK fuzzy system is proposed to identify seizures from EEG signals. Functional brain networks of epileptics are constructed by analyzing the synchronization between multi-channel EEG signals. Complex network theory is further applied to extract topological features of brain networks. Taken the network parameters as independent inputs, a fuzzy system based TSK model is established and further trained through supervised learning to identify seizures. Experimental results demonstrate the effectiveness of the proposed scheme. The accuracy, sensitivity and specificity of seizure states identification are 98.36%, 99.48%, and 97.24%, respectively. The novel method, combining complex network theory with machine learning algorithm, provides a potential tool for identifying epilepsy state from EEG signals, which has important values in clinical application.

Key words: EEG, complex network, fuzzy system, machine learning

摘要： 脑电检测是癫痫疾病诊断的重要手段，但基于脑电信号特征的人工标记方法，对癫痫发作状态识别的准确度较低。将脑功能网络与TSK模糊系统相结合，提出一种癫痫脑电信号识别的新方法。通过分析多通道脑电信号之间的同步性，构建癫痫患者的脑功能网络，采用复杂网络方法提取特征参数；以脑网络参数为输入特征建立TSK模糊系统模型，通过监督式学习训练分类器，用于识别癫痫发作期的脑电波形。实验结果证明了该方法的有效性，模糊分类器对癫痫发作状态识别的准确度达到98.36%，99.48%敏感度和97.24%特异度。该方法将复杂网络与机器学习算法相融合，为通过脑电检测识别癫痫疾病状态提供了新方法，具有重要的应用价值。

关键词: 脑电信号, 复杂网络, 模糊系统, 机器学习