基于无创电磁高频振荡脑网络的难治性局灶性癫痫致痫区定位研究

Precise localization of epileptogenic zone is the key point to treat intractable focal epilepsy by neuromodulation and surgical resection. Massive researches show that high frequency oscillation (HFOs) is an underlying biomarker of epileptogenic zone, and several kinds of noninvasive techniques can record HFOs. But the range of brain tissue generating HFOs is much wider than epileptogenic zone. Existing methods to localize epileptogenic zone in the basis of HFOs cannot be applied for clinic due to the unsatisfactory accuracy. Therefore, more effective method to localize epileptogenic zone according to HFOs is needed. This research will use noninvasive electromagnetic technique based on scalp electroencephalogram, magnetoencephalogram and neuronal current magnetic resonance imaging to record interictal and ictal HFOs, set up the multi-dimension HFOs networks by Granger causality analysis, identify the relationship between the key nodes and connections of HFOs brain networks and the epileptogenic zone by deep learning method, and develop technique in view of noninvasive electromagnetic HFOs network to accurately localize epileptogenic zone. Besides, the causal relationship between the change of HFOs network after surgery and seizure termination will be established to explore the brain network mechanism of seizure termination. This research will provide a new noninvasive method to localize epileptogenic zone in network level, as well as provide new thought to research the mechanism of seizure termination.

精准定位致痫区是应用神经调控和手术切除方法治疗难治性局灶性癫痫的关键环节。大量研究显示高频振荡是癫痫致痫区的潜在生物标记物，并且多种无创方法可以记录到高频振荡。但产生高频振荡的脑组织范围通常大于致痫区范围，已有的根据高频振荡特征定位致痫区的方法的准确性尚未达到应用于临床的标准，因此需要研发更有效的基于高频振荡的定位致痫区的新方法。本研究将应用基于头皮脑电图、脑磁图和神经元电流磁共振成像的无创电磁技术，记录难治性局灶性癫痫患者发作间期和发作期的高频振荡，利用格兰杰因果分析建立多维度高频振荡网络，应用深度学习方法得到网络中的重要节点和连接与致痫区的关系，发展基于无创高频振荡脑网络的精准定位致痫区的新方法，然后建立手术前后高频振荡脑网络的改变与癫痫发作终止的因果关系，探索癫痫发作终止的脑网络机制。本研究将在脑网络层面上为精准定位致痫区提供新的无创方法，还将为研究癫痫发作终止的机制提供新的思路。

精准定位致痫区是应用手术及神经调控治疗难治性局灶性癫痫的关键环节。大量研究显示高频振荡是癫痫致痫区的潜在生物标记物，并且多种无创方法可以在成人癫痫患者中记录到高频振荡。但目前尚无有效的基于无创高频振荡的定位致痫区的方法，并且术后癫痫发作消失的网络机制尚未明确。本研究利用高密度头皮脑电图、脑磁图记录了难治性局灶性癫痫患者发作间期高频振荡，发现无创高频振荡具有较强定位癫痫致痫区的价值。利用相干和定向传递函数方法建立多维度高频网络，与正常人对比后发现，癫痫患者高频网络中显著连接的节点和区域与致痫区存在显著相关性，提示病理性网络连接的存在。我们还分析了手术预后良好（Engel 1级）的患者手术前后脑网络的改变，发现术前显著连接的区域术后相干系数显著减少，提示病理性网络连接的中断可能为术后癫痫发作消失的机制之一。本研究建立了记录和分析无创高频振荡的技术，发展了基于无创高频脑网络的精准定位致痫区的新方法，探索了手术后癫痫发作消失的可能脑网络机制，在高频网络层面上提供了癫痫治疗的新思路。