【作者】 门聪；

【导师】 王江；

【作者基本信息】 天津大学 ， 检测技术与自动化装置， 2012， 博士

【摘要】 神经系统以时间空间编码的形式表征外部刺激所包含的信息，但神经系统对于针刺等外部刺激编码解码以及神经电信号传输通路建模的研究鲜见报道。本论文一方面根据针刺传输通路的生理结构建立针刺前馈网络信号传输模型并研究其动力学特性，另一方面基于实验数据从编码、解码、复杂网络等角度挖掘针刺神经信号，并与模型分析的结果对照。为揭示针刺作用机理，本文根据针刺神经信息传输通路的生理结构建立了针刺信号传输通路的前馈网络模型。根据进一步假设把针刺作用等效为噪声，发现针刺对调节网络编码特性（放电率和放电规则性）具有重要作用。噪声下前馈网络发生了相干共振现象，表明一定强度的针刺可能引起网络共振和同步，从而改变神经元网络的特性，进而产生针刺效应。把针刺刺激等效成伪周期信号，发现耦合神经元网络的相干共振能够由异质性非周期信号诱导产生，通过改变异质性非周期信号的平均幅值、周期或者异质性可以改变网络相干共振的特性。模型研究表明网络共振和同步是针刺发生作用的可能机理。在针刺信号通路模型分析的基础上，本文设计了不同手法和频率的针刺作用大鼠足三里的动物实验，获取了脊髓背根神经电信号，根据放电类选算法提取针刺神经信号编码特征，首次发现不同手法所对应的时空编码形式存在本质的差别，尤其是捻转法与提插法的编码形式差异较大；这种差异性体现在神经元对特定针刺手法的选择性，但对同一手法不同频率的针刺作用，这种编码选择性并不明显；还发现了神经系统对于针刺作用具有适应性和饱和特性。在此基础上，利用贝叶斯解码算法通过编码分析的结果反向预测手法，预测成功率接近百分之九十。通过统计分析，证明所得结果具有统计意义，并使用互信息量化了针刺信息。这些结果可以直接应用于神经系统和机器接口的实现。本文首次提出采用复杂网络映射的方法分析针刺神经电信号，证明捻转法和提插法两种针刺手法的内在特征是不同的。本文对实验数据的特征和针刺前馈网络模型的特征进行对比，发现针刺前馈网络可以复现针刺现象，证明了前馈网络模型的合理性。实验和模型研究的结合对进一步理解针刺和提高临床治疗的效率具有非凡的意义。更多还原

【Abstract】 Neural system characterizes information of external stimulation by spa-tiotemporal encoding. However, there are rarely reports about the encodingform of acupuncture stimulation and the model of electrical signal transmissionpath. We try to explore the underlying mechanism of acupuncture by construct-ing and analyzing the feedforward neuronal model and mining the experimentalsignals based on encoding, decoding algorithm and complex network mappingmethod.First, we constructed a feedforward network (FFN) of FitzHugh-Nagumo(FHN)neurons according to the feedforward pathway of acupuncture signal. Here, theacupuncture stimulation is equivalent to noise. It is found that acupuncture playsan important role in modulating the transmission of fring rate and spiking reg-ularity. Furthermore, noise could induce coherent resonance in the feedforwardnetwork. Therefore, acupuncture with certain intensity can induced resonance inneuronal network, which would afect the activities of the neural system. More-over, after assuming the acupuncture stimuli to be aperiodic signal, it is foundthat resonance can also be induced by heterogenous aperiodic signals, and pa-rameters of the aperiodic signal can modify the resonance of the coupled network.These studies imply that resonance and synchronization may be the mechanismof acupuncture.To further understand the mechanism of acupuncture, experiments are de-signed that manual acupuncture (MA) manipulations with diferent types andfrequencies are taken at ’Zusanli’ points of experiment rats. We found the spa-tiotemporal coding pattern of diferent acupuncture manipulations. The difer-ences between features of manipulation ’nb’’nx’ and those of manipulation ’tb’’tx’ are obvious. However, neuronal selectivity of encoding is not obvious whenthe same manipulation is taken with diferent frequencies. Neuronal adaptivityand saturation phenomenon are also observed when acupuncture with diferentfrequencies are taken. Types of acupuncture manipulations taken on the rats are inferred witha high probability (about90%) by Bayesian decoding algorithm based on theresponse of multiple neurons. These results are proved to be signifcant by s-tatistical analysis. Furthermore, mutual information is applied to quantify thedecoding process. These studies may help to construct the interface betweenneural systems and machines and improve the clinical study. We also proposecomplex network mapping method to analyze the signal. It indicates that theinherent characteristics of MA ’nb’’nx’ and those of MA ’tb’’tx’ are diferent.After compare the experimental results and the model analysis, we foundthat some key properties in experiment can be reproduced by the feedforwardmodel. This fnding imply that the modelling strategy is reasonable. The combi-nation of experimental and computational study is helpful to improve the clinicaltreatment of acupuncture.更多还原