QRS detection of ECG signal using U-Net and DBSCAN

Abstract

QRS detection is a crucial task for ECG signal analysis, which is the preliminary and essential step to further recognition and diagnosis. This paper proposes a U-Net based method for QRS detection. The method consists of three steps including preprocessing, U-Net model, and density-based spatial clustering of applications with noise(DBSCAN). The normalization is carried out using the Z-score method in preprocessing. In this study, location prediction is conducted by the U-Net model. Subsequently, the U-Net outputs are thresholded and clustered by DBSCAN. Finally, the middle points of the cluster are regards as the R-peak of the QRS complex. We demonstrate that the proposed method achieving high accuracy on ECG signals from the MIT-BIH Arrhythmia Database(MITDB). The experimental results show an average sensitivity of 99.98 %, positive predictivity of 99.95 %, accuracy of 99.93 %, and F1-score of 99.97 %. Compared with other existing methods, the overall performance is comparable and even better in terms of accuracy and F1-score.