• DocumentCode
    473738
  • Title

    Cardiac model based approach to QT estimation

  • Author

    Povinelli, RJ ; Mneimneh, MA ; Johnson, MT

  • Author_Institution
    Marquette Univ., Marquette, MI
  • fYear
    2006
  • fDate
    17-20 Sept. 2006
  • Firstpage
    333
  • Lastpage
    336
  • Abstract
    Because the direction of regulatory agencies requires the study of potential proarrhythmic side effects of drugs be conducted by skilled readers, the 2006 Physio- net/Computers in Cardiology Challenge focuses on the need for clinically acceptable automatic measurement of the QT duration. This entry to the challenge approaches the problem from an inverse ECG perspective. The model parameters are estimated through the minimization of the squared error between the generated ECG signal from the activity of the AV node, endocardium and epicardium, and the recorded ECG. The PQ junction and the end of the T wave are determined from the activation and deactivation of the ventricular activity. This approach is tested on the PTB Diagnostic ECG Database, which contains 549 records from 294 subjects. The final score for this method was substantially improved from 130.70 ms to 51.69 ms as compared to the best score of 17.33 ms for automatic methods.
  • Keywords
    drugs; electrocardiography; least squares approximations; 2006 Physionet-Computers in Cardiology Challenge; AV node activity; PQ junction; PTB Diagnostic ECG Database; QT duration automatic measurement; QT estimation; T wave end; cardiac model; drug proarrhythmic side effects; endocardium activity; epicardium activity; inverse ECG; squared error minimization; time 130.70 ms; time 51.69 ms; ventricular activation; ventricular deactivation; Cardiology; Clinical trials; Computer errors; Databases; Drugs; Electrocardiography; Heart rate interval; Parameter estimation; Signal generators; Testing;
  • fLanguage
    English
  • Publisher
    ieee
  • Conference_Titel
    Computers in Cardiology, 2006
  • Conference_Location
    Valencia
  • Print_ISBN
    978-1-4244-2532-7
  • Type

    conf

  • Filename
    4511856