Research On Detection And Classification Algorithm Of Arrhythmia Based On Convolutional Neural Network

Arrhythmia is a common cardiovascular disease syndrome,and the detection of arrhythmia has positive significance for the prevention of cardiovascular disease.Electrocardiogram is a graphical technology that reflects the dynamic process of depolarization and repolarization of cardiomyocytes.By observing whether the waveform changes of electrocardiogram are abnormal,you can warn and diagnose cardiovascular diseases.In clinical examination,the acquisition of ECG signals is susceptible to noise signals such as power frequency noise,EMG interference,and baseline drift,which leads to certain difficulties in correctly identifying arrhythmia.In addition,the waveform of arrhythmia is more complicated,and the huge number of electrocardiograms will dramatically increase the work pressure of cardiologists.Therefore,the method of relying only on the doctor's experience to identify the type of arrhythmia is inefficient,and there is a subjective possibility of misdiagnosis and missed diagnosis.This paper is based on Convolution Neural Network(CNN),which mainly studies the preprocessing method of ECG signal and the detection and classification algorithm of arrhythmia,and uses the MIT-BIH database to verify the algorithm.Finally,we compare and analyze the experimental results.This paper mainly studies from the following three aspects:1.Research on noise filtering of original ECG signal.First,the effects of wavelet transform and morphological filtering methods in the baseline drift noise filtering process are introduced,and their respective defects are analyzed.Secondly,a new method for filtering baseline drift noise by combining wavelet transform and morphological filtering is proposed.Finally,the comparative analysis results show that the new method has a more significant effect in suppressing baseline drift in ECG signals.2.Research on the detectionc and lassification algorithm of arrhythmia.First of all,the QRS wave detection and heartbeat interception methods are used to extract the arrhythmia data,and the traditional classification technology of arrhythmia detection is introduced.Secondly,a classification model for arrhythmia detection based on onedimensional convolutional neural network is proposed,which combines feature extraction and detection classification.In the end,using the MIT-BIH database data for model experiment verification,through model parameter optimization,the network quickly converges,and achieves an effective classification effect of five types of heartbeats.3.Analyzes and researches algorithm experiment results.First,introduce the evaluation indicators of the classifier.Then configure and optimize the parameters of the number of convolution kernels and the number of network layers,and analyze the results of the heart beat data mixing matrix,the detection classification accuracy of the model reaches 99.10%.Finally,Finally,the classification results of this method are compared with the classical SVM classification method and artificial feature classification method.The results show that this method is better than the traditional classification method,it skips the steps of feature extraction and selection,avoids the model's excessive dependence on the accuracy of feature extraction,and reduces the complexity of the calculation process caused by feature transformation.