Ecg Automatic Analysis And Diagnostic Approach

Electrocardiographic signal can be used to analysis and distinguish all kinds of arrhythmia,so the technology of electrocardiographic signal automatic analysis has very high value on clinic,and it is one of the top focuses which the present domestic and international scholar studies.Because the signal is very complex and nonlinear,which increases the degree of difficulty on analysis,and the automatic diagnosed effect can not achieve the specialist' on computer. So the researcher are still, persistent to improve the traditional methods and explore new resolutions. The article deep studies as follows:1. The denoising of cardiogram signal.At first,we introduce several common kinds of methods of filter design with the purpose of removing the interference of Baseline wander,Powerline and Eletromyography,and design the filter based on wavelet transform ,which and increase the veracity of classification.2. The feature extraction of electrocardiographic signal.After compared with several common kinds of methods of feature extraction, the article adopts the multi-resolution to decompose the cardiogram signal with four scales with the mallat algorithm, and detects the singularity on it, and extract feature with exact position.3. The automatic classification of arrhythmia. It introduces the current methods of pattern recognition and artificial neural network at first, the article adopts the methods of BP neural network and probability neural network to classificate at four arrhythmia and one normal with the classification performance of 97.62% on BP.4. the automatic diagnosis ot atrial fibrillation. The first step is extracting features,then calculate the Euclidean distance and make the decision with K-neighbors finally.the test's result indicates that the sensitivity is 90.0% and the specificity is 87.5% and the validity is 78.9%.