脑深部电刺激（Deep brain stimulation, DBS）作为临床应用最广泛的神经调控技术之一，在全球老龄化问题日益严重的现实下，具有重要的临床价值。然而，人们至今对其调控机制的认识并不完全。主要原因是人脑的复杂性和特殊性，导致现有研究缺乏直接的、长期的电生理或生物影像数据，来进一步说明DBS究竟如何恢复了受损大脑的功能。清华大学神经调控技术国家工程实验室经过长期努力，成功研发具有同步记录功能的DBS系统。该系统通过对DBS靶点——基底节的局部场电位进行监测，为揭示DBS的长期调控作用提供了有效的研究工具。这是本文研究工作的基础。基于这一技术，本文从电生理的角度出发，针对DBS的长期调控作用开展了深入的研究工作。首先，本文针对长期、植入、同步电刺激这一特殊的信号采集条件，明确了LFP信号中各类噪声的来源、特点以及去除方法。其中，重点分析了同步采集研究中常遇到的亚谐波成分，通过实验验证了其噪声的本质，完善了获取有效的局部场电位的方法。进而，利用具有同步采集功能的DBS设备，对清醒状态的帕金森病患者基底节局部场电位进行术后1、3、6和12个月的跟踪记录，研究了DBS对病理性Beta节律的调控，结果显示相同刺激强度下的抑制作用随时间显著降低。通过建立模型，清楚地解释了DBS电生理调控作用与临床改善程度之间的关系，并明确了DBS长期调控方式由高Beta频段向低Beta频段迁移的规律，表明DBS对低Beta频段的调控可能是长期的起效机制。同时，辅助同步睡眠监测手段，对睡眠状态帕金森病患者基底节局部场电位进行跟踪记录，研究了不同睡眠阶段下的节律特征，验证了其在长期范围内的一致规律，明确了DBS在不同睡眠阶段下的长期调控作用。最后，本文系统性地验证了上述基底节的节律特征是清醒/睡眠状态长期、可靠的生物标记。在此基础上，成功研发了全植入式闭环DBS系统，并初步验证了临床可行性，推动了闭环DBS设备走向临床应用。
As one of the most widely applied surgical neuromodulation techniques in the brain, deep brain stimulation (DBS) holds extraordinary medical and clinical potential, especially against the growing issue of social aging across the globe. However, the mechanisms of DBS are still not well understood. The primary reason for this is the lack of long-term direct electrophysiological or biological image data on the exact effects of DBS treatment in the impaired brain. Currently, an advanced DBS device with sensing function of local field potentials (LFPs) from the surgical target is successfully developed by National Engineering Laboratory for Neuromodulation of Tsinghua Universtiy, which is the foundamental of this thesis. Based on this technology, long-term effects of DBS were studied through the LFPs recorded from the subthalamic nucleus of the basal ganglia. Firstly, with focus on the long-term implanted LFP sampling process that was mixed with strong artifacts of electrical pulses, we conducted a study to clarify the influence of signal quality from various factors. With electrical circuit simulation and in-vitro tests, the quality of LFP signal was proved to be influenced by the ripples modulated in the DBS pulses, noise from the switches, other electrophysiological signals (such as electrocardiogram), and long-term effects of implantation. A series of methods were built to purify the effective LFPs.Secondly, the DBS sensing device was used to trace the long-term LFP signals in the basal ganglia of patients with Parkinson’s disease. The results revealed that the beta suppression of DBS was consistent over the long-term. However, the suppression decreased after long-term DBS. A machine learning model was successively built between the changing beta suppression and the corresponding clinical improvement with DBS. From the analysis of the model structure, the long-term modulation pattern with the most weights shifted from high beta band to low beta band.Together with synchronized sleep monitoring, LFPs in the subthalamic nucleus of the patients with Parkinson’s disease were recorded. The different sleep patterns of the LFP rhythm was clarified. In parallel, effects of DBS during different sleep stages were proved. The sleep patterns and the DBS modulation effects during sleep were proved to be consistent over long term monitoring. Finally, based on the results from our studies, we summarized a long-term reliable biomarker of wakefulness and sleep stages based on subthalamic nucleus LFPs. The results show that the characteristic band may change from the high frequency band to the low frequency band. A new fully implantable closed-loop DBS system based on the algorithms for the classification between wakefulness and sleep was developed. The primary clinical feasibility was proven and it is fundamental for future clinical applications.
