Research On Feature Extraction Of ECG And Its Application

With the acceleration of people's life rhythm and the increase of work pressure,the trend prevalence and death rates grown fast for cardiovascular diseases and have become a major threat of human life and health.Electrocardiogram(ECG),as a comprehensive manifestation of human heart activity,is of great significance in the detection and diagnosis of cardiovascular diseases.There are many methods for analyzing ECG signals,but the effect of ECG signal analysis methods for mobile monitoring is often not ideal.The use of 12-lead electrocardiographs in the market requires specialized medical personnel to operate and cannot be applied to ordinary personal homes.In view of the above problems,this dissertation studies a variety of ECG signal processing methods,and based on analyzing the characteristics of noise,proposes a set of ECG signal detection algorithms.The algorithm includes two parts: ECG signal preprocessing algorithm and feature parameter extraction algorithm.The ECG signal preprocessing methods include filtering,five-point derivative method,normalization square and moving window integration.In the feature parameter extraction algorithm,this dissertation proposes an R-peak positioning algorithm called derivative crossover method,which including the two times of first-order derivative method,compensation scheme and cross-differentiation.Based on this method,the feature parameters of QRS complex,J-point and ST interval in ECG signal are also extracted.In order to test the advantages and disadvantages of the derivative crossover method,the R peak location algorithm dynamic threshold method and the derivative crossover method were compared experimentally using simulated data and clinical data.Simulation results show that the accuracy and sensitivity of the derivative crossover method are both 99.75%;the accuracy and sensitivity of the dynamic threshold method are both 99.65%.This dissertation uses the records in the MIT/BIH database to compare the two algorithms.The results show that the accuracy of the derivative crossover method is 98.24% which is higher than the dynamic threshold method rate of 87.47%,and the sensitivity of the derivative crossover method is 99.46% which is higher than the dynamic threshold method rate of 98.95%.The above experiments show that the derivative crossover method proposed in this dissertation is an algorithm with good detection effect and has a certain use value.In order to improve the data transmission efficiency,this article also designed a method of ECG data encoding,compression and decompression.Through calculation,this method can compress the data to 52.59%.Based on the above research results,this article designed a real-time ECG detection system for detecting arrhythmia.The system consists of three parts,namely the acquisition module,Cortex platform and terminal equipment.The acquisition module collects the ECG data and transmits it to the Cortex platform.The platform processes the ECG signal and transmits the detected information to the terminal equipment.The terminal equipment provides an interface display platform.The system can realize long-term detection of the patient's ECG signal,and discover abnormal heart activity in time to provide data support for clinical diagnosis.