Ecg signal analysis, diagnosis and transmission

Abstract

The electrocardiogram (ECG or EKG) signal plays an essential role in the field of medical science and is used in the diagnosis of various cardiovascular diseases (CVD). If the recorded ECG signal has an irregularity in the heartbeat rhythm, it is known as arrhythmia. The diagnosis is done based on ECG morphology. Heartbeat can be subdivided into five categories such as normal, supraventricular ectopic, ventricular ectopic, fusion and unknown beats. In this paper, the authors present an easy and effective way for analysing and diagnosing the nature of the arrhythmia using 1D convolutional neural network (CNN). The ECG data set was obtained from PhysioNet’s MIT-BIH database. The PyTorch library was used in python in designing the CNN model which classifies the ECG test sample signal to a category. The diagnosed signal along with the patient details is sent to a cardio specialist for validation via mail and to the CVD forum android application where the case history can be maintained in the database and can be reviewed by any cardio specialist. The CNN model was trained using the data sets and achieved an average accuracy of 97.72%, and the classification achieved is presented in a confusion matrix.