Surgical resection of the epileptogenic zone (EZ) is an effective method for treating drug-resistant epilepsy. At present, the accuracy of EZ localization needs to be further improved. The characteristics of graph theory based on partial directed coherence networks have been applied to the localization of EZ, but the application of network control theory to effective networks to locate EZ is rarely reported. In this study, the method of partial directed coherence analysis was utilized to construct the time-varying effective brain networks of stereo-electroencephalography (SEEG) signals from 20 seizures in 12 patients. Combined with graph theory and network control theory, the differences in network characteristics between epileptogenic and non-epileptogenic zones during seizures were analyzed. We also used dung beetle optimized support vector machine classification model to evaluate the localization effect of EZ based on brain network characteristics of graph theory and controllability. The results showed that the classification of the average controllability feature was the best, and the area under the receiver operating characteristic (ROC) curve (AUC) was 0.9505, which is 1.32 % and 1.97 % higher than the traditional methods. The AUC value increased to 0.9607 after integrating the average controllability with other features. This study proved the effectiveness of controllability characteristic in identifying the EZ and provided a theoretical basis for the clinical application of network controllability in the EZ.

中文翻译：

---

基于时变有效网络的致痫区定位


手术切除致痫区（EZ）是治疗耐药性癫痫的有效方法。目前EZ定位精度有待进一步提高。基于部分有向相干网络的图论特性已被应用于EZ定位，但将网络控制理论应用于有效网络来定位EZ的报道却很少。在这项研究中，利用部分定向相干分析的方法构建了 12 名患者 20 次癫痫发作的立体脑电图 (SEEG) 信号的时变有效脑网络。结合图论和网络控制理论，分析癫痫发作时致癫痫区和非致癫痫区网络特征的差异。我们还利用粪甲虫优化的支持向量机分类模型来评估基于图论和可控性的脑网络特征的EZ的定位效果。结果表明，平均可控性特征分类效果最好，受试者工作特征（ROC）曲线下面积（AUC）为0.9505，比传统方法提高了1.32%和1.97%。将平均可控性与其他特征相结合后，AUC值增加至0.9607。该研究证明了可控性特征在识别EZ中的有效性，为网络可控性在EZ中的临床应用提供了理论基础。