如果能够在对帕金森病（Parkinson’s disease, PD）患者进行脑深部电刺激（deep brain stimulation, DBS）治疗的同时，获取长期的神经电活动信息，将为研究DBS作用机制、实现真正的闭环刺激提供一个难得的机遇，进而有可能为研究脑功能疾病建立一个基于人脑的研究平台，加深我们对大脑的认识。本文首先研究同步电刺激下深脑局部场电位（local field potential, LFP）采集方法与装置，并搭建体外测试平台。针对由于电极触点间距小、刺激电压高所导致的同步刺激下采集信号中包含幅值高于有效信号两到三个数量级的刺激伪迹使得信号频谱失真的问题，研究全频带信号重建的方法。首先提出了基于模板建立与提取的方法，即建立刺激伪迹的通用模板并从采集数据中减去从而重建有用信号；第二种方法是应用现代信号处理中自适应参数估计的理论，将刺激伪迹当成一系列正弦信号的叠加，估计它们的频率、相位和幅值，再从原始信号中减去并得到期望的全频带有用信号，该方法潜在地为闭环DBS的实时处理提供技术支撑。在此基础上，记录批量PD患者术中丘脑底核（subthalamic nucleus, STN）LFP信号，初步探索了PD患者STN-LFP节律特征和DBS对其影响规律，发现：① 与多巴胺能药物类似，DBS可以抑制β频段过多的同步活动，但未观察到类似于药物引起的γ频段振荡，表明DBS治疗机制可能不同于药物疗法；②在1~3V刺激幅度范围内，β频段振荡的抑制程度与刺激幅度成正相关，90μs脉宽表现出比60μs对β振荡更大的抑制；在60~120Hz刺激频率范围内，β频段的抑制程度与刺激频率成正相关，在120~185Hz范围内，β频段的抑制随刺激频率的增加无明显变化趋势；该研究为临床刺激参数优化提供了技术支撑；③在11/75例刺激关闭的STN-LFP中发现高频振荡（260±62Hz），17/75例STN-LFP在刺激开启时存在快速高频振荡（300~350Hz），猜想快速高频振荡可能是促进运动的节律，电刺激可能会引起fHFO，该现象为DBS机制的研究提供了线索。在上述装置、处理方法研究和临床数据验证的基础上，研制了可实时传输出深脑LFP的全植入系统，即带感知功能的脑起搏器系统。采用该系统，以PD食蟹猴模型为对象的长期全植入验证性动物实验表明发病侧肢体PD症状评分值和对侧STN-LFP的β频段电活动水平均随造模天数呈递增趋势，二者归一化值高度相关。最后完成了首例临床植入试验，系统功能正常，成功采集到患者深脑LFP信息。
The technological capability for obtaining neural activity in patients with Parkinson’s disease (PD) during deep brain stimulation (DBS) treatment will provide a valuable opportunity for elucidating the mechanisms of therapeutic action, particularly closed-loop DBS. Such research tools have the potential for being developed and expanded for exploring other major brain disorders.The research carried in this thesis demonstrates the development of an implantable device for the local field potential recording (LFPs) recording during simultaneous DBS. In current configurations, the relative distance between electrodes (four) combined with large amplitude electrical impulses during DBS leads to stimulation artefacts, of which are approximately 2-3 orders of magnitude larger than LFPs occupied by electrode resulting in power spectrum distortion. In the first study, methods for removing DBS artefacts were developed using an in vitro platform. We first applied stimulation artefact building and subtraction in order to maintain the full-band physiological signal in LFPs recorded during DBS. A second method for removal of artefacts was carried out by isolating stimulation artefacts as a combined series of sinusoids, also known as harmonics. Frequencies, amplitudes and phases of sinusoids were estimated adaptively and subtracted from recorded LFPs during DBS-ON to obtain the desired signal. These data suggested this method can be potentially achieved in the implanted neurostimulator allowing technical support for revealing real-time closed-loop control in DBS. Next, a clinical trial was conducted to study the pathological synchronisation of neurons in the subthalamic nucleus of PD patients during DBS therapy. We found that during DBS, there was a suppression of β-band oscillations without the predominance of γ-band oscillations, indicating a different therapeutic mechanism of action to pharmacotherapy. We next investigated the relationship of the β-band synchronisation variations and applied DBS parameters in the STN. The results demonstrated the degree of β-band power attenuation was elevated following increased voltage in a range of 1~3V, and 90μs stimulation showed more suppression of β band power than 60μs stimulation. The β power decreases with increasing stimulation frequency within a range of 60 to 120 Hz, but didn’t present an obvious declining trend anymore when the stimulation frequency was programmed to the clinically commonly used high frequency range which is about 120 to 185Hz. This study can potentially provide a clue for the DBS parameter optimization clinically. From the data collected in our clinical trial, we found that 17 of 75 STN-LFPs in ON DBS presented fast high frequency oscillations (fHFO, 300-350 Hz), while 11 STN-LFPs in OFF DBS displayed slow HFO (260±62Hz). From these findings, we hypothesise that fHFO may be a prokinetic rhythm, and DBS may cause fHFO, providing a clue in the mechanism of DBS. Following studies on hardware development, data analysis and clinical trials, an implantable DBS system with the ability of signal sensing and real-time transmission was designed for testing in preclinical research. Using the PD monkey model, the implanted device allowed recordings of β-band activity in the right STN, which were increased with the symptomatic scores of the left limb over time. These normalised values were highly correlated with each other. Thereafter, this device was implanted in a patient with PD and recordings and readouts of STN-LFPs have so far been successful.
