Automated Prediction of Spontaneous Termination of Atrial Fibrillation from Electrocardiograms

Dieter Hayn, Kurt Edegger, Daniel Scherr, Peter Lercher, Brigitte Rotmann, Wolfgang Klein, Günter Schreier

    Research output: Contribution to journalArticlepeer-review

    Abstract

    An algorithm for differentiating ECGs with atrial fibrillation (AF) that will spontaneously terminate within 60 seconds from signals, where it won't, has been developed using the AF termination challenge database from physionet. The algorithm was based on the calculation of the major AF frequency by canceling out the QRS complexes and T waves from the original ECGs and then applying short time Fourier transform techniques to the remaining signals. The major AF frequency and the mean RR interval were considered for classification. Validation of the algorithm was done by sending the algorithm's results for test-set-a of the AF termination challenge database to physionet. We found, that for ECGs with a low AF frequency it was more likely, that AF would terminate spontaneously than for ECGs with higher frequencies. Our algorithm was able to correctly classify 93.3% (28/30) of the signals of the test-set-a.
    Original languageEnglish
    Pages (from-to)117-120
    Number of pages4
    JournalComputers in Cardiology
    Volume31
    DOIs
    Publication statusPublished - 11 Jul 2005
    EventComputers in Cardiology 2004 - Chicago, United States
    Duration: 19 Sept 200422 Sept 2004

    Research Field

    • Not defined

    Keywords

    • Atrial fibrillation
    • Electrocardiography
    • Frequency
    • Testing
    • Cardiology
    • Spatial databases
    • Telemedicine
    • Signal processing
    • Fourier transforms
    • Drugs

    Fingerprint

    Dive into the research topics of 'Automated Prediction of Spontaneous Termination of Atrial Fibrillation from Electrocardiograms'. Together they form a unique fingerprint.

    Cite this