Rule-based Method For Morphology Classification Algorithm Of ST-T Segment In ECG Signals

According to the World Health Organization, cardiovascular diseases(CVD) are one cause of death worldwide. A new research shows that the number of death caused by CVD is only less than the number of death caused by gross tumor. Currently, with the development of technology and growth in living standing, long time working lead to more and more people getting CVD. The World Health Organization calls for people to pay more attention to the prevention of CVD, not only the treatment of CVD. Myocardial ischemia is one of the CVD and the incidence is rising, especially for the old people. Our country gradually enters the aging time, myocardial ischemia prevalence is increasing, with the limitation of health resources and medical level, the health system under tremendous pressure in our country.ECG monitors are common devices in hospital and family, they can record the changes in real-time ECG signals. With the development of information technology, the portable ECG monitors are in the ECG monitoring market. For example, the mobile ECG monitoring, smart watches. For the ECG is weak signal, the mobile collection of ECG is in strong background noise, and the number of data is huge. It is depend on the artificial recognition before diagnosis, which lead to the impossible of mobile diagnosis alarm. Besides, data storage need to consume memory, the cost of the products improved. For wireless transmission, a large amount of data means that consume large amounts of traffic also can cause network congestion. It needs to have a preprocessing about the ECG signals before the data transfers to the server. A reliable and efficient ECG processing algorithm is the hotspot for the researchers, and also the development of mobile ECG monitoring bottlenecks. Compared to the research of QRS waveforms, less research on the ST-T segment morphology changes about myocardial ischemia is relatively small. This study proposed the method for ST-T segment morphology classification, the contents are shown below:(1) Signal preprocessing and feature points extraction. The features points extraction have a great influence on the classification accuracy. In this study, the author determines the feature point extraction algorithm by contrasting several methods.(2) ST segment morphology classification. First, cardiologists give annotations for ST segment morphology, after signal processing and feature points extraction, this study designs the classification algorithm corresponding to the annotations and use training data to optimize algorithm and testing data to verify algorithm.(3) T wave morphology classification. First, cardiologists classify T wave morphology into 5 types. After preprocessing and extract T wave signals, this study designs classification algorithm and parameters according to the T wave morphology. T wave signals are also divided into training set to optimize algorithm and testing set to verify algorithm.