Research On Extractions Of Arteriovenous Pathological Characters And Methods Of Multi-dimensional Nonlinear Analysis

Hypertension and other cardiovascular diseases have become the most and most serious diseases endangering human in contemporary. Therefore, it is necessary and urgent to enhance treatment, monitoring and prevention of these diseases. Presently, these diseases'non-invasive detection mainly relies on monitoring of ECG, heart sound and arterial pulse, which are one-dimensional qualitative analyses. The relationship between different organs of human is complex and impacts on each other. So, this simple signal from a single full in-depth qualitative analysis is difficult to reflect systemic problems. The arterio-venous wave signal has characteristics of non-stationary, non-Gaussian, non-linear. But the traditional frequency domain and time domain method is only based on a simple linear fitting and statistical analysis. Since arteriogram and venous signals usually contain information about the conditions of biosystem and biostructure and that they are weak signals comparing to strong background noise, they have the fectures of non-Gaussianity, nonlinearity and nonstationarity, so it's useful to research this kind of system by using time-frequency analysis. In order to get a better analysis results, multi-dimensional non-linear methods should be adopted.Based on domestic and abroad signal analyses and research of arteriogram and venous wave, this dissertation gives the detailed time-frequency analysis of arteriogram and venous signals, and presents the advantages and disadvantages of time domain analysis, frequency domain analysis and time-frequency analysis. Due to the features of the arteriogram and venous signal, which are nonperiodicity, nonstationarity and analogous-periodicity and etc, this dissertation mainly focuses on the implementation of locating the specified point of arteriogram and venous signals. Firstly, definitions and characteristics of time-frequency analysis are introduced in the dissertation. The advantages and disadvantages of time-frequency analysis are dicussed, and some typical time-frequency methods are presented. The model and specified point of a normal arteriogram and venous signal are defined, and the relationships of the typical arteriogram and venous signal are analyzed. Then with the aid of software, such as MATLAB, etc, the typical arteriogram and venous signal is researched and analyzed by the use of wavelet denoise and time-frequency analysis. In connection with the signality of locating the specified point, this paper raises the method of using the instantaneous frequency to locate it exactly with the aid of MATLAB.At the present, Wigner-Ville and Choi-Willams analyses are the mostly used time-frequency analysis methods. The merits of Wigner-Ville analysis is the best Coefficientofregression, fine verge characteristic and local energy collecting, while the shortcoming of it is serious distracter. Choi-Willams analysis is called as smooth pseudo-Wigner-Ville analysis. Choosing the suitable coefficient of Cohen type can reduce the effection of cross term. Finally, this dissertation uses the improved Choi-Willams analysis to process the arteriogram and venous signals. The results from theoretic analyses and simulations show that this method is useful to reduce the effection of cross term and to extract the potential information, and that it can be used to process actual arteriogram and venous signals.