The Research And Application Of Intelligent Electrocardiogram Diagnosis System

Cardiovascular disease is one of major diseases leading to human death.It has the characteristics of concealment,suddenness and high mortality,and seriously harms human health and life safety.With the gradual acceleration of population aging in China and the increasing attention paid by people to heart health,the demand for daily ECG monitoring is increasing.However,the existing portable ECG monitoring equipment,with single function,disconnected data,limited storage capacity and weak analytical and processing capacity,cannot really play a role in helping understand health status and instantly discover cardiovascular diseases.In this context,this paper aims to achieve an efficient and reliable ECG intelligent diagnosis system to help people prevent and detect potential cardiovascular diseases.On the one hand,this paper designs a set of accurate and effective ECG automatic classification methods,including ECG signal denoising,feature point detection,waveform feature extraction and pattern recognition algorithm of arrhythmia.On the other hand,the hardware module of ECG acquisition and the web-based ECG monitoring network platform are implemented concretely.The specific research content is as follows:(1)research on pretreatment of ECG signals.ECG signals are weak and easily affected by the environment in the acquisition process.In order to improve the signal-to-noise ratio of signals,this paper designs an ECG pretreatment method combining wavelet denoising and traditional filters to eliminate interference noises such as power-line interferences and baseline drift in ECG signals.(2)A QRS detection algorithm based on Hilbert transform and adaptive threshold method was proposed.A differential operation and Hilbert transform are applied on the denoised signal in order to suppress the effects of P and T waves and decrease noise like baseline drift.Then,adaptive threshold method and error check mechanism were used to detect the envelope signal peak.The algorithm was verified in MIT-BIH arrhythmia database,and the detection result of QRS wave reached 99.01%.(3)To develop arrhythmia classification algorithm.This paper proposes and compares the classification model based on the parallel combination of LSTM and CNN,as well as the arrhythmia classification algorithm combining features extraction methods such as PCA,ICA and wavelet packet entropy with classifiers such as random forest.The experimental results show that the former achieves better accuracy.(4)The development of ECG intelligent diagnosis system,including the design of ECG acquisition hardware equipment and ECG monitoring network platform.The former realizes ECG signal acquisition,conditioning and transmission,while the latter provides users with web-based medical services,including personal information management,automatic classification of ECG,telemedicine and other functions.