ECG Quality Assessment Using 1D-Convolutional Neural Network

Abstract

Recently many wearable devices have been developed for casual health management. ECG is an important tool for monitoring of cardiac arrhythmias. Wearable devices are able to collect electrocardiogram (ECG) data in a convenient and low-cost way. ECG should have sufficient quality for ECG experts to analyze and make a clinical decision. Identifying low quality signals in advance will significantly facilitate clinical diagnosis. In this paper, we use 1D convolutional neural network (CNN) to classify single-lead ECG signals with a duration of 10 s as "acceptable" or "unacceptable". Through validating with the database of PhysioNet/Computing in Cardiology Challenge 2011 and the PhysioNet/Computing in Cardiology Challenge 2017, the accuracy of the proposed method is 94.3%.