• DocumentCode
    541706
  • Title

    Automatic electrocardiogram delineator based on the Phasor Transform of single lead recordings

  • Author

    Martínez, A. ; Alcaraz, R. ; Rieta, J.J.

  • Author_Institution
    Innovation in Bioeng. Res. Group, Univ. of Castilla-La Mancha, Cuenca, Spain
  • fYear
    2010
  • fDate
    26-29 Sept. 2010
  • Firstpage
    987
  • Lastpage
    990
  • Abstract
    The present work introduces a new ECG delineator, based on the Phasor Transform, which is able to operate in single lead recordings. The method converts each instantaneous ECG sample into a phasor, thus being able to deal very precisely with P and T waves, which are of notably lower amplitude than the QRS complex. Initially, the method relies on the detection of R peaks and, next, onset and offset of the QRS complex are identified. Finally, taking the QRS as a reference, P and T waves are detected and delineated. This delineator was validated with the QT database, available at Physionet, providing average values of sensitivity higher than 98.60% for the detection of all the significant ECG waves and fiducial points. Concretely, P wave sensitivity was 98.65% for the onset, peak and offset. The QRS onset and offset achieved a sensitivity of 99.85% and, finally, the T wave provided a sensitivity of 99.20% both for its peak and offset. Additionally, the average maximum time delineation error was lower than 6 ms and its standard deviation was in agreement with the accepted tolerances for expert physicians in the onset and offset identification for QRS, P and T waves. As a consequence, this new algorithm is able to achieve a performance similar to the top rated ECG delineators, but with notably lower computational cost.
  • Keywords
    electrocardiography; medical signal detection; medical signal processing; ECG; P waves; Physionet; QRS complex; QT database; R peaks; T waves; automatic electrocardiogram delineator; phasor transform; single lead recordings; Cardiology; Databases; Electrocardiography; Robustness; Sensitivity; Wavelet transforms;
  • fLanguage
    English
  • Publisher
    ieee
  • Conference_Titel
    Computing in Cardiology, 2010
  • Conference_Location
    Belfast
  • ISSN
    0276-6547
  • Print_ISBN
    978-1-4244-7318-2
  • Type

    conf

  • Filename
    5738141