The Researh And Application Of Feature Matching Algorithms Forecg Data Classification

Cardiovascular disease is currently one of the highest morbidity and mortality diseases, The automatic analysis of ambulatory electrocardiogram(ECG) is very important measure to diagnosis cardiovascular diseases. which is very effective of treating and preventing cardiovascular diseases, and the occurrence of unexpected situations will decrease. The main purpose of ECG analysis are ECG waveform identification and classification. The current ECG classification algorithms are mainly neural network, fuzzy classification, feature extraction, template matching, support vector machine clustering algorithm and so on.The main purpose of this paper is to introduce a combining electrocardiogram(ECG) classification algorithm based on feature extraction and template matching. The principle of the algorithm is based on morphological characteristics between normal waveform and abnormal waveform, combining waveform RR interval and QRS complex width to classify. The purpose of the algorithm is to identify premature ventricular contractions(PVC) and supraventricular premature contraction from a long ECG data, which are two common arrhythmia. In this paper, the MIT-BIH database is used to evaluate algorithm performance, achieving a higher classification accuracy. the paper firstly introduces the basics of ECG and authoritative arrhythmia database(MIT-BIH Database and AHA Database) used for evaluating algorithm performance. Then the paper implements the feature matching ECG classification algorithm. The paper introduces the details of the algorithm implementation process, including signal preprocessing, QRS complex wave detection, template matching and so on. Finally, the MIT-BIH database is used to evaluate the algorithm performance, and analyze the statistical results.The ECG feature matching algorithm is applied to the classification system based on dynamical characteristics and network topology characteristics, This paper introduces the details of the feature matching algorithm implementation process, and comparing with a single feature extraction algorithm. This paper uses the database for evaluating the algorithm performance, achieving a good performance. The paper also analyzes the deficiency of previous algorithms.The proposed ECG classification algorithm not only rapidly and accurately analyze ECG but also make classification statistics for two typical heart diseases. Medical personnel can be free from a large number of ECG analysis with focus on two kinds of abnormal ECG data, The method can greatly improve the efficiency of medical diagnosis.