The Research On Dynamic Pulse Signal Detection And Real-time Pulse Rate Variability Extraction And Analysis

Pulse signal and pulse rate variability(PRV) signal contain abundant physiological and pathological information of cardiovascular system, are often employed to prevent, diagnose and cure some cardiovascular disease(CVD). Because pulse signal is easy to obtain, it has been used in many portable medical equipments. Detecting dynamic pulse signal, extracting and analyzing PRV signal, are important for monitoring and forewarning the CVD in real time.In this work, first, the physiological mechanism and the clinical value of pulse signal and PRV signal are summarized, and the research status of pulse signal detection systems, PRV extraction methods, and their analysis approaches are illustrated. In addition, for the noise and interference suppression and signal quality assessment during pulse signal acquisition, and the contradiction between the accuracy and real-time capability of the existing method in PRV extraction and analysis, some methods are proposed to preprocess dynamic pulse signal and to extract and analyze dynamic PRV(DPRV) signal. The main contents of this thesis are:1) For some noises and interference are easy to be filtered in pulse signal, a method named self-adjustable parameters integer-coefficients filter is proposed to preprocess the pulse signal under different signal to noise ratio(SNR). On the basis of integer-coefficients filter, the signal smooth degree is defined to evaluate the filtering results and adjust the parameters. The proposed method is applied to the simulated and the actual pulse signals, the experimental results show that the proposed method can more quickly and more efficiently suppress some noise and interference in dynamic pulse signal compared with other filter methods.2) For some artifacts are not wiped out or suppressed in pulse signal, an artifacts detecting and signal quality method is proposed. The artifacts are divided into pulse artifacts, no-pulse signal segments and motion artifacts by their characteristics, the corresponding detection methods are proposed and denoted as artifacts cl assifying and detection method, which are employed to detect the artifacts from the pulse signals of the international authoritative database and the pulse signal are acquired by ourselves. The results show that the proposed method can quickly extract the artifacts from pulse signals. Moreover, from the results of artifacts, the quality of pulse signal are assessed, and the low quality signals are removed. Only the high quality signal s are retained.3) According to the characteristics of pulse signal in tim e domain and frequency domain, and the merit and demerit of common PRV extraction methods, there are three PRV extraction methods are proposed: adaptive threshold method, the methods based on improved sliding window iterative DFT(ISWIDFT) and Hilbert-Huang transform(HHT), which are employed to analyze the simulated and the measured pulse signals. The results show that, the ISWIDFT(fundamental component) method is more quickly and more accurately to extract PRV signal than other method. Meanwhile, it is not influenced by the changes of SNR and sampling frequency, and can be used to DPRV signal in real time.4) For the defects such as bad real-time performance, great computational complexity of the nonlinear and the time domain PRV analysis methods, the sliding window iterative theory is used to improve these methods, which are used to detect CAD. The improved methods are as follows: methods in time domain, Poincare plot analysis, base-scale entropy analysis(BSEA) and sign series entropy analysis(SSEA). The experimental results show that the improved methods can quickly analyze PRV signal after updating sample, and are efficient to analyze DPRV in real time. In addition, some parameters are obtained with the improved methods. A characteristic vector consists of these parameters are used to classify the younger and the older, and the healthy subjects and the coronary heart disease(CHD) subjects from international authoritative database with mechine leaning methods, the classification results show a high level of accuracy and the improved method can be used to detect some CAD.5) A system is designed for detecting and processing dynamic pulse signal based on smart mobile platform. By detecting and processing actual dynamic pulse signal, It can be infered that the self-adjustable parameters integer-coefficients filter, the signal quality assessment method based on artifacts classifying and detection are useful for dynamic pulse signal detection, and the ISWIDFT(fundamental component) method and the improved PRV analysis methods are efficient to extract and analyze DPRV signal in real time, and derive some parameters from DPRV signal. These methods are efficient for monitoring and forewarning the CVD.