New ECG Delineation Database

In this paper, the UNNCyberHeart Database (UNNDB) of ECG signals is presented for validation of ECG signals delineation algorithms. The database consists of 200 10-second 12-lead ECG signal records representing a wide class of different morphologies of the ECG signal. The boundaries of the ECG signal complexes were manually annotated by cardiologists for all 200 records independently for each lead. In addition, all records have a complex diagnosis. With this database a new algorithm for delineation of the ECG signal was successfully tested, taking into account the set of ECG signal leads and automatically determining the morphologies of the complexes. Earlier this algorithm showed competitive accuracy on QTDB database, which has only 2 leads. All the advantages of the developed algorithm were completely revealed when analyzing the fully annotated 12-lead database presented in this paper.