This database is described in
Please cite this publication when referencing this material, and also include the standard citation for PhysioNet:
In addition, when referencing the atr annotations of this database, please acknowledge their creators (MEDICALgorithmics Ltd, Warsaw, Poland; efforts coordinated by Michal Tadeusiak).
This database includes 84 long-term ECG recordings of subjects with paroxysmal or sustained atrial fibrillation (AF). Each record contains two simultaneously recorded ECG signals digitized at 128 Hz with 12-bit resolution over a 20 mV range; record durations vary but are typically 24 to 25 hours.
Two sets of annotations are available here:
- The qrs annotations were produced by an automated QRS detector; in these annotation files, all detected beats (including occasional ventricular ectopic beats) are labelled N, detected artifacts are labelled '|', and AF terminations are labelled T.
- The atr annotations were obtained by manual review of the output of an automated ECG analysis system (see below); in these annotation files, all detected beats are labelled by type, and rhythm changes are also annotated (see the summary tables for details).
The AF Termination Challenge Database consists of 80 one-minute excerpts of a subset of these records (those numbered 00 through 75). As a guide to selecting the excerpts, “T” were inserted manually in the qrs annotation files of this subset of the records, to mark spontaneous terminations of AF episodes with durations of at least one minute.
The original recordings were collected and contributed to PhysioNet by Steven Swiryn and his colleagues at Northwestern University. They were digitized and automatically annotated at Boston’s Beth Israel Deaconess Medical Center. Steven Swiryn and George Moody annotated the AF terminations.
The atr annotations were developed on PhysioNetWorks and contributed to PhysioNet by MEDICALgorithmics Ltd (Warsaw, Poland). The recordings have been automatically annotated using MEDICALgorithmics’ PocketECG system algorithm (FDA 510(k): K112921) and manually verified by an experienced team of MEDICALgorithmics' ECG technicians. The reference annotation development project was coordinated by Michal Tadeusiak.
Thanks to Mariano Llamedo Soria for reporting an error in the original version of 20.hea, and for providing a correction incorporated in the current version.
Segments from this database were used in: