Research On Algorithm For ECG Beat Classification With Abnormal Rhythm Analysis

Cardiovascular disease has become the first killer to human health, how to effectively diagnose the disease and decrease morbidity and mortality has become one of public health problem which is urgent to solve. ECG(electrocardiogram) which direct recorded the regular change of waek current during heart beating,plays an unsubstitute role in cardiovascular diseases diagnosis with the advantages of non-invasive,rapid and accurate.How to analyse ECG automatically by computer has important significance for improving clinical diagnostic efficiency and accuracy,and has attracted a wide spread attention.Based on previous work,this thesis introduced abnormal ECG rhythm analysis, combining with the beat template classification and morphologic parameter analysis to classify three types ECG beat automatically.The main works in this thesis are as follows:First,the thesis introduces the research status of ECG automated anlysis technology on three aspects:siganl pretreatment,wavefrom detection and feature parameter extraction,arrhythmia automaic dignosis.Then it describe some existing algorithms and discuss the clinical application status and prospect briefly.Second,it detects the abnormal beat by using beat template classification base on the result of QRS detection and then determines some feature parameter threshold base on morphologic feature parameter extraction.Third,the thesis detects abnormal rhythm by analyzing heart rate characteristics and special waveforms and realized three different rhythm(Atrial fibrillation, Atrial flutter, Supraventricular tachyarrhythmia) auomatic analysis.Fourth, base on the works above, it classify three types ECG beat (Premature ventricular contraction, Atrial premature beat, Normal beat) automatically by logic judgment. Then MIT-BIH arrhythmia database is used to evaluate the classification result. Finally, the thesis designs and implements the ECG algorithm analysis platform and ECG automatic analysis system software.Experimental results show that the algorithm is effective to classify three types ECG beat, the sensitivity of PVC and APB classification was reached 96.9% and 99.7% respectively. The beat misjudgments were significantly decreased by introducing abnormal rhythm analysis and the efficiency and reliability of the result of classification was improved.