Studies On Early Detection Of Myocardial Ischemia Via Deterministic Learning And Its C++ Implementation

Myocardial ischemia is a relatively common cardiovascular disease, a serious threat to people’s life and health. In our country, cardiovascular disease has become the leading cause of death among urban and rural residents. Even more worrying is that cardiovascular disease patients are gradually getting younger and younger. Therefore, it is of great significance to study the early detection of myocardial ischemia by ECG.In the past ten years, Prof. Wang Cong group has carried out a series of innovative research in the field of dynamical patterns modeling and identification, and put forward a new theory of artificial intelligence- deterministic learning theory. And ECG signals are essentially a kind of temporal or dynamical patterns generated from electrical activities of the heart. Therefore, the research group attempted to use deterministic learning for the early detection of myocardial ischemia, developed the prototype based on Matlab, and proposed a new method for the early detection of myocardial ischemia- Cardiodynamicsgram(CDG). Matlab has a natural advantage in large-scale matrix computation, which provides a great convenience for the early prototype development. But Matlab has great limitations in engineering and commercial applications. C++, as the most commonly used application development language, has excellent performance in high performance computing. This thesis focuses on the study of prototype implementation using C++. The main works of the thesis are as follows:1. ECG data acquisition and pre-processing. In order to shield the complexity of Windows API functions and data types, this thesis encapsulates a simple serial communication class CSerialPort. Then developing the program of upper-computer to achieve the original ECG signals acquisition based on the class. And further by data converting, median filtering and wavelet threshold filtering, the original signals’ noises interference are effectively removed, the routine 12-lead ECG are obtained. This provides an important guarantee for the further realization of ECG accurate dynamical modeling and effective feature extraction.2. Fast implementation of deterministic learning algorithm by Armadillo+OpenBLAS platform. C++ does not support linear algebra, which is undoubtedly a challenge for algorithm implementation. Armadillo as an open source linear algebra library has a high degree of similarity with the Matlab syntax and high scalability. It can be used in conjunction with high performance basic linear algebra subprograms OpenBLAS, in some performance even beyond Matlab. This provides a new thought for the implementation of deterministic learning algorithm by C++. In the thesis, based on the improved computing solutions and program optimization, the fast dynamical modeling of ECG signals is realized by Armadillo+OpenBLAS platform. It will not only solve the engineering problems, but also provide a great convenience for the commercial.3. Based on the acquisition of ECG data and the realization of the deterministic learning algorithm, this thesis builds an early detection system of myocardial ischemia using MFC framework. It mainly consists of three parts, including the ECG signals acquisition subsystem, the analysis subsystem and the 3D visualization display. Then the CDG method and its effectiveness are introduced.In the thesis, an early detection system for myocardial ischemia was developed based on C++, which effectively solved the problem of strongly dependence, low efficiency and poor confidentiality of the Matlab prototype. Also laid a good foundation for further promotion of the CDG method and carrying out clinical trials.