| In recent years, cardiovascular disease has become the leading cause of death in our country. Studies have shown that, the factors causing cardiovascular disease have gradually increased with the improvement of people’s living standards, accompanied with the population aging and population growth, the morbidity and mortality of cardiovascular disease is still in the rising phase in our country. According to statistics, the current number of Chinese patients with cardiovascular disease has reached 290 million people, so early diagnosis and treatment of cardiovascular disease is particularly important.With the advances in medical technology, many new technologies have been proposed, which can be used to diagnose coronary artery disease in cardiovascular disease. Such as coronary artery angiography(CAG), intravenous ultrasound (IVUS), optical coherence tomography (OCT) etc. However, both the CAG, which has been the "gold standard" of coronary artery disease diagnosis, and the IVUS,which has a high guidance value on interventional diagnosis and treatment of coronary artery disease, can only offer imaging evaluation on vascular lesions, rather than directly assess the physiological functions of the coronary artery disease site accurately. Doctors can only employ their diagnosis experience to determine the nature of coronary artery stenosis, which may not accurately evaluate the severity of the disease, hence causes serious consequences. In order to evaluate the physiologic functional status of coronary stenosis site more accurately, domestic and foreign researchers always are looking for new and effective evaluation method.1993 Pijls, Vanson and other scholars presented a new concept named coronary fractional flow reserve (FFR):FFR is defined as the ratio of the maximum the myocardium blood flow with the presence of stenosis and the maximum blood flow assuming the absence of stenosis at the same region. Compared with the previous method of imaging evaluation, coronary FFR can accurately evaluate the severity of coronary artery stenosis, which will not be affected by hemodynamic factors (such as blood pressure, heart rate, myocardial contractility, etc.) theoretically.At present, coronary FFR device is divided into two parts, one is the pressure guide wire; Second is the measurement host. The measurement process is:pressure guidewire enters the bloodstream from a human lower limb arteries or lumbar arteries with the guidance of angiographic technology to arrive at the coronary stenosis and measure the blood pressure values to both ends of the stenosis by the ultra-miniature pressure sensors which located at distal end of pressure guidewire, then send them to measurement host to perform signal processing, and calculate FFR value. Pressure guide wire acts as the core element FFR measurements, its measurement accuracy will directly affect the FFR measurement results. However, the ultra-miniature pressure sensor located at the distal end of pressure guide wire has the extremely small size, and has inherent non-linear problem affected by manufacturing process. And it is also affected by ambient temperature seriously. Since the desired blood pressure for calculation of FFR is measured by ultra-miniature pressure sensors directly, its nonlinearity and temperature drift will directly affect the FFR measurement accuracy. Therefore, we have to eliminate the nonlinear and temperature drift of ultra-miniature pressure sensor, find an effective compensation method to reduce its nonlinear and temperature error to ensure the accuracy of the FFR measurement results.Aiming at the problems of pressure sensor nonlinear and temperature compensation, there are manly hardware compensation and software compensation. The current FFR measurement systems used clinically are primarily made by the United States St. Jude and Volcano, and both companies use the hardware method to compensate the temperature and nonlinear error of the pressure guide wire. Our previous members have done the hardware compensation method research. To avoid hardware compensation methods patent protection which have been declared by foreign companies, more importantly, to further improve the compensation accuracy, the software compensation method with low cost and high compensation accuracy is needed to realize nonlinear and temperature compensation of pressure guide wire. But there is no pressure guide wire temperature and nonlinear error software compensation research now, as for the traditional pressure sensor temperature and nonlinear software compensation, the artificial intelligence algorithms are relatively mature, especially BP neural network algorithm. BP neural network can approximate nonlinear mapping model in any accuracy by training, and improve the compensation accuracy effectively, but the standard BP neural network has the shortcomings, such as easy to falling into local optimum, low convergence speed and poor stability etc. For the shortcomings and deficiencies of the prior technologies, this paper proposes an improved particle swarm algorithm to optimize the weight and threshold of neural network to improve the BP neural network compensation effect, and apply it in the pressure guidewire temperature and nonlinear error compensation. Through experiments, compared to the hardware compensation method, the use of software algorithms for pressure guidewire temperature and nonlinear error compensation reduces the compensation costs and improves the compensation accuracy. While using the global optimization capability of improved PSO algorithm to optimize BP neural network can avoid it falling into local optimum and improve its compensation accuracy generalization ability and stability. Finally, the paper takes preliminary FPGA hardware implementation to the pressure guidewire compensation model by improved algorithm, its compensation result is basically same with Matlab simulations.First The paper introduces the background of our research, including the definition of fractional flow reserve, the preliminary work our group has done and problems in our preliminary work, and describes the development of domestic and international pressure guide wire temperature and nonlinear error compensation and the related technical background of FPGA, and discusses the content and significance of this research in detail, and describes the structure arrangement of the paper at last.The second chapter of the paper describes the working principle of the pressure guide wire, specifically describes its temperature and nonlinearity errors, then complies the hardware compensation method and analysis its compensation effect, also introduce the software compensation principles and methods of conventional silicon pressure sensor temperature and nonlinearity errors. After studying and analyzing the present various software compensation methods in-depth, which have been widely used, combining the specific application environment of software compensation method, we select the software compensation method of improved PSO optimizing BP neural network to compensate pressure guide wire Nonlinear and temperature error.Then, the paper introduces the BP neural network algorithm and particle swarm optimization. First, it describes the structure of the neural network model, common transfer functions, and specific training model in detail. As one of the most widely used artificial neural network algorithms, BP neural network has a lot of advantages on pattern recognition, pattern matching, pattern classification and decision support, etc. especially on construction of nonlinear systems. We also have done a specific analysis of BP neural network advantages and disadvantages on the temperature and nonlinearity compensation of the pressure guide wire. After comprehensively comparing swarm intelligence algorithms, selecting the particle swarm optimization to optimize BP neural network, which has the advantages that high convergence speed and strong global search capability and the algorithm is easy to realize and so on, and then specifically describes the particle swarm algorithm work principles and processes, at the same time we make some improvement to the deficiencies of PSO on the global optimization process based on the current research of other scholars.The fourth chapter of the paper describes pressure guidewire temperature and nonlinear error compensation simulation and implementation based on the improved PSO-BP Algorithm. First, we design the pressure guide wire sample data acquisition systems, the hardware part of the lower machine in the data acquisition systems uses STC12C5A6S2 as the microcontroller core, uses PGA309 to complete amplification and zero offset compensation, uses ADS1115 to complete A/D-conversion, and communicate with PC via RS232 serial port. The display and storage interface of pressure guidewire pressure and temperature signals on position machine part is designed by LABVIEW to display the wave and value of pressure signal and the value of temperature signal. Then it introduces the Matlab simulation of the improved PSO-BP algorithm, and it specifically introduces the software processes of the BP neural network compensation optimized by improved particle swarm optimization; and it also discusses the nonlinear system implementation scheme of pressure guidewire inverse measurement model by FPGA.At last, we analyze the effects of pressure guidewire temperature and nonlinear error compensation by the improved PSO-BP algorithm and the implementation of software compensation by FPGA. After the comparative analysis of the compensation effects of standard PSO-BP algorithm, non-optimized BP neural network algorithm and the improved algorithm, we can summarize that the improved PSO-BP algorithm has advantages that high compensation accuracy, strong generalization ability and high stability etc. While the advantages that the improved PSO-BP network algorithm has a simple structure, less parameters, little amount of computation, etc. making it possible to compensate pressure guidewire temperature and nonlinearity error of FFR measurement system in high precision, high stability and low cost. |
PDF Full Text Request