Jin Seung-Hyun, Chung Chun Kee
Neuroscience Research Institute, Seoul National University College of Medicine, Seoul, Republic of Korea; iMediSyn Inc., Seoul, Republic of Korea.
Neuroscience Research Institute, Seoul National University College of Medicine, Seoul, Republic of Korea; Department of Neurosurgery, Seoul National University Hospital, Seoul, Republic of Korea; Department of Neurosurgery, Seoul National University College of Medicine, Seoul, Republic of Korea; Department of Brain and Cognitive Sciences, Seoul National University College of Natural Sciences, Seoul, Republic of Korea.
Epilepsy Res. 2017 Jan;129:138-145. doi: 10.1016/j.eplepsyres.2016.11.018. Epub 2016 Nov 23.
The main aim of the present study was to evaluate whether resting-state functional connectivity of magnetoencephalography (MEG) signals can differentiate patients with mesial temporal lobe epilepsy (MTLE) from healthy controls (HC) and can differentiate between right and left MTLE as a diagnostic biomarker. To this end, a support vector machine (SVM) method among various machine learning algorithms was employed. We compared resting-state functional networks between 46 MTLE (right MTLE=23; left MTLE=23) patients with histologically proven HS who were free of seizure after surgery, and 46 HC. The optimal SVM group classifier distinguished MTLE patients with a mean accuracy of 95.1% (sensitivity=95.8%; specificity=94.3%). Increased connectivity including the right posterior cingulate gyrus and decreased connectivity including at least one sensory-related resting-state network were key features reflecting the differences between MTLE patients and HC. The optimal SVM model distinguished between right and left MTLE patients with a mean accuracy of 76.2% (sensitivity=76.0%; specificity=76.5%). We showed the potential of electrophysiological resting-state functional connectivity, which reflects brain network reorganization in MTLE patients, as a possible diagnostic biomarker to differentiate MTLE patients from HC and differentiate between right and left MTLE patients.
本研究的主要目的是评估脑磁图（MEG）信号的静息态功能连接是否能够区分内侧颞叶癫痫（MTLE）患者与健康对照（HC），并能否作为一种诊断生物标志物区分右侧和左侧MTLE。为此，在各种机器学习算法中采用了支持向量机（SVM）方法。我们比较了46例经组织学证实患有海马硬化且术后无癫痫发作的MTLE患者（右侧MTLE = 23例；左侧MTLE = 23例）与46例HC的静息态功能网络。最优的SVM组分类器区分MTLE患者的平均准确率为95.1%（敏感性 = 95.8%；特异性 = 94.3%）。包括右侧后扣带回在内的连接性增加以及至少一个与感觉相关的静息态网络的连接性降低是反映MTLE患者与HC之间差异的关键特征。最优的SVM模型区分右侧和左侧MTLE患者的平均准确率为76.2%（敏感性 = 76.0%；特异性 = 76.5%）。我们展示了电生理静息态功能连接的潜力，其反映了MTLE患者的脑网络重组，作为一种可能的诊断生物标志物可用于区分MTLE患者与HC以及区分右侧和左侧MTLE患者。
