【作者】 刘新文；

【导师】 王惠南；

【作者基本信息】 南京航空航天大学 ， 精密仪器及机械， 2006， 博士

【摘要】 基于电生理信号的靶点定位是立体定向脑神经外科手术的关键技术,本文根据临床需要,以手术中微电极记录的帕金森病人神经细胞放电为对象,以获取放电信号特征为主线,以提取核团位置关联因子为目标,实现对微电极在神经核团中的位置实时定位,达到精确引导手术的目的,解决临床术中定位难题。本文的主要研究内容是信号预处理;放电脉冲增强、检测;脉冲提取、分类;特征因子求取;微电极术中定位。经过深入系统的研究,给出了详细的处理方法和步骤,对一些关键性技术取得了如下创新性成果。1、提出了小波变换去噪预处理方法,找出了适合于人脑神经放电的小波基和最小分解层数。剔除了神经放电信号的基线漂移和高频噪声干扰,避免了信号处理引起的放电波形失真。2、提出了小波系数和非线性能量算子两种放电增强算法,提出了神经放电脉冲检测和脉冲提取分步独立实现的方法。对于信号弱、信噪比低的神经放电信号,经处理提高了信噪比,克服了放电检测准确性和放电波形失真的矛盾。3、提出了自适应阈值新的检测算法,给出了两种检测阈值算法、四种放电检测方法。提取出了不同核团的脑神经放电脉冲,并对神经放电脉冲进行了分类,找出了人的神经放电持续时间及其规律。利用模拟神经放电验证了检测方法的准确性,采用临床神经放电信号验证了其可靠性。4、找出了与位置关联的盒子维、检测阈值、能量调制、峰峰间隔四个位置特征因子,并给出了具体算法。提出了利用四种因子实现定位的方法,在脑神经外科立体定向手术中首次提出了合理的定位依据和客观的定位技术,同时还给出了神经放电信号的可视化定位方法。5、提出了关联维无标度区中心自动选取的方法,实现了关联维的自动计算,解决了无标度区中心选取需要人工干预的难题。6、设计并构建了动物实验系统,采集了大鼠的脑神经放电信号,验证了放电脉冲提取和四种特征因子定位方法的可靠性。更多还原

【Abstract】 Electrophysiological target localization is the key technology in microelectrode guided stereotactic neurosurgery for Parkinson’s disease, MER-based functional targeting locate the nominal anatomical target location based on MRI more precisely. The research is based on clinic microelelctrode data of Parkinson’s disease. The goal of the study is to identify neuronal structures of microelectrode depth intraoperatively by extracting the characteristics of neuronal discharge signals; to realize objective and automated precise real time localization and to achieve microelectrode guided surgery by analyzing characteristics of the MER data. An in-depth study is given systemically of signal pretreatment, spike enhancement, spike detection, spike extraction, spike sort techniques and characteristic parameters, key technologies of locating microelectrode position, methods of signal processing is given step by step. The main achievements and innovative results are described in the following parts:1、A wavelet transform based method is introduced for removing the baseline draft and high frequency noises in neuronal signals at the same time. The wavelet type for human neuronal spike and optimal decomposition scale levels are selected for avoiding discharge wave distortion during denosing process.2、Two enhancement algorithms of wavelet coefficients and nonlinear energy operator are applied to improving signal to noise ratio. The process of spike detection and extraction are proposed separately to solve the contradiction between detecting accurately and wave distortion of neuronal spikes.3、An exact detection is foundation of spike characteristic analysis. The detection precision is influenced by detection threshold . Two algorithms of threshold and four detection methods are introduced to extract neuronal spikes of different nucleus. A novel algorithm of adaptive threshold detection is presented , the threshold is adjusted automatically for various amplitude and signal to noise ratio. These algorithms are verified by simulative signal and clinical MER signal. The methods of spike detection and classification are correct. Neuronal spikes are extracted of different nucleus, the latency of human neuronal discharge is obtained. Spikes are classificated using PCA.4、Four feature parameters of box dimension, detection threshold, energy modulation, interspike interval are extracted from microelectrode recordings for objective and quantitative target localization, four targeting techniques are presented. The subjective localization of microelectrode position intraoperatively is overcomed. Four targeting techniques can improve localization accuracy, and can be used in clinic neurosurgery. Visual targeting method is given in addition.5、A novel method of identification the fractal scaleless center in correlative dimension is proposed. The choice of fractal scaleless is unsupervised based on the formula presented, correlative dimension can be calculated automatically.6、The animal experimental system is set up. The neuronal spike data are sampled by microelectrode, and processed by spike extraction.four targeting parameters is ued for microelectrode localization. Methods of spike extraction and four targeting techniques are validated by animal experiment.更多还原