Muzhitskaya N. Express-diagnosis based on discrete wavelet analysis of pulse waves

The thesis describe the development of methods and tools for express diagnosis based on discrete wavelet analysis of pulse waves. A comparative analysis of methods and software of modern automatic processing of pulse waves is done. A harmonic three-phase model of pulse wave is designed. It is supplemented by concrete numerical indicators for brachial artery. Specified set of hemodynamic parameters, based on the new model proposed, these parameters considers as a basic for express-diagnosis procedure. Proved the advantages of discrete wavelet analysis in the problem of selection of informative component of the pulse wave, proved its high accuracy, the ability to work in real time, suitability for long-term storage of output results. The method of processing of model signals proposed, as a result recommended basic wavelets, the number of decomposition levels and method for synthesis for purified signal allows you to get correlation coefficient up to 0.9995 with the original signal. The proposed method of synthesis of pulse signal from coefficients of the discrete wavelet decomposition was more effective in comparison with popular threshold methods. The structure and principles of the automatically pulse wave processing systems, that based on the established methods, developed. Implemented an experimental model and studied the real pulse waves, resulting in a confirmed efficiency of established methods.