Abstract
Electrocardiograms (ECGs) are widely used for diagnosing cardiac arrhythmia based on the deformation of signal shapes due to changes in various heart diseases. However, these abnormal signs may not be observed in some of the 12 ECG channels, depending on the location or shape of the heart and the type of cardiac arrhythmia. Therefore, to accurately diagnose cardiac arrhythmias, it is necessary to closely and comprehensively observe ECG signals acquired from 12 channel electrodes. In this study, we proposed a clustering algorithm that can classify persistent cardiac arrhythmia as well as episodic cardiac arrhythmias using the standard 12-lead ECG signals and the 2D CNN model using the time-frequency feature maps to classify the eight types of arrhythmias including normal sinus rhythm. The standard 12-lead ECG dataset was provided by Computing in Cardiology 2020 Physionet Challenge and consisted of 6,877 patients. The proposed algorithm showed excellent performance in the classification of persistent cardiac arrhythmias; however, its accuracy was somewhat low in the classification of episodic arrhythmias. If our proposed model is trained and verified using more clinical data, we believe it can be used as an auxiliary device for diagnosing cardiac arrhythmias.
Time–Frequency Feature Fusion for Noise Robust Audio Event Classification
