【作者】 孙力；

【导师】 王江；

【作者基本信息】 天津大学 ， 检测技术与自动化装置， 2008， 博士

【摘要】 针灸是一种传统的治疗方法,探索针灸的功能及其机理,对临床诊断及研究有重要的指导价值。研究神经信号编码的机制,是当前神经科学领域中的重要研究内容。随着混沌理论的发展,它已经在包括生物医学工程在内的众多领域获得了广泛的应用。信号的时频多尺度分析与信息论也是信号处理中的研究热点。本文采用现代时间序列分析方法对针刺电信号进行了深入研究并得出以下结论。本文设计两组实验,分别为针刺足三里测取大鼠脊髓背根神经束动作电位、针刺足三里测取大鼠脊髓背角神经元动作电位。采用了峰峰间期的概念,对针刺电信号进行一阶回归映射动力学结构分析以及概率分布拟合统计学分析,区分了不同手法针刺电信号,初步证明了针刺电信号是混沌的;将针刺电信号看做随机点过程,研究各种手法之间在时域与频域的相关性,得到了其中两种手法相似度很高,其它手法不相关的结论。首次采用非线性动力学方法对四种不同手法针刺电信号进行了分析。首先得到了电信号的功率谱,证明其是混沌的。利用互信息法确定时间延迟,Cao法确定最佳嵌入维数,对针刺电信号进行相空间重构,得到手法的吸引子;在相空间重构基础上,得到了吸引子的特征量:用GP算法求关联维数,用Wolf法求最大Lyapunov指数,并求解了时间序列的Kolmogorov复杂度。采用递归图来确定针刺电信号的平稳性,采用替代数据法来判别信号的非线性确定性特性。最后基于相空间重构的加权零阶局域预测法对针刺电信号进行预测。上述分析方法证明了针刺电信号具有混沌特征。将小波理论与信息论方法结合,对针刺电信号进行分析。建立了基于小波分析的小波能量熵、小波时间熵、小波奇异熵和小波时频熵的概念与计算方法。应用小波熵于针刺电信号的特征提取与分类上,能有效地将不同手法区分开来。本论文的结论为针灸的量化及科学化奠定了一定的理论基础。更多还原

【Abstract】 Acupuncture is a systemic science; thus to study its function and mechanism provides guiding meaning for clinical diagnosis and research. It’s an important research part in neuroscience to explore the mechanisms of the neural coding. With the development of chaos theory, it has been applied in numerous areas including biomedical engineering. Meanwhile, the signal time-frequency analysis and information theory are also popular research methods.In this thesis, we design two kinds of experiments. One is acupuncture on Zusanli to obtain action potentials on the spinal dorsal root, the other is acupuncture on Zusanli to obtain action potentials on the spinal dorsal horn.We obtain the interspike-interval (ISIs) of the acupuncture signals evoked by different manipulates, and then use the first return map and probability distribution fit to analyze them. By considering the concept of the point process, correlation analysis in time domain and frequency domain is employed to analyze the spike trains.The nonlinear time series analysis methods are proposed to study four acupuncture electrical signals for the first time. Power spectrum analysis is applied to imply the signals are chaotic. Then phase space reconstruction is applied among which mutual information method to obtain the time delay and Cao’s method to obtain the embedding dimension, and stranger attractors are reconstructed. On the basis of the reconstruction, we characterize the attractors in terms of correlation dimensions, largest Lyapunov exponents and Kolmogorov complexity measures. Two additional methods are chosen: recurrence plot to determine the stationarity of the signals and the surrogate data method to estimate the deterministic nonlinearity property of the signal. The weighted zero-order local prediction method is applied to forecast the electrical signals.Last but not the least, by combining the wavelet theory and information theory, fundamental definitions of wavelet entropy measure are discussed. Calculation methods including wavelet energy entropy, wavelet time entropy, wavelet singular entropy, wavelet time frequency entropy are put forward to extract the characteristics of the electrical signals, and their meanings are analyzed.The results of this thesis can give some theoretical supports to the quantification and scientification of the acupuncture.更多还原