Download PDF
Methods Inf Med 1975; 14(04): 202-207
DOI: 10.1055/s-0038-1635711
Original Article
Schattauer GmbH

Optimal Values for the Parameters of an On-line Algorithm Monitoring the QRS Waveform[*)]

Optlmale Parameterwerte Eines On-Line-Algorithmus Zur Überwachung Des Qrs-Komplexes
Ch. van Eyll
1   Cardiovascular Laboratory, the Coronary Care Unit and the Laboratory for Experimental Surgery, University of Louvain, Belgium
,
W. Strepenne
1   Cardiovascular Laboratory, the Coronary Care Unit and the Laboratory for Experimental Surgery, University of Louvain, Belgium
,
J. Lefevre
1   Cardiovascular Laboratory, the Coronary Care Unit and the Laboratory for Experimental Surgery, University of Louvain, Belgium
,
J.-L. Bachy
1   Cardiovascular Laboratory, the Coronary Care Unit and the Laboratory for Experimental Surgery, University of Louvain, Belgium
,
J. Col
1   Cardiovascular Laboratory, the Coronary Care Unit and the Laboratory for Experimental Surgery, University of Louvain, Belgium
,
J. Cosyns
1   Cardiovascular Laboratory, the Coronary Care Unit and the Laboratory for Experimental Surgery, University of Louvain, Belgium
,
A. A. Charlier
1   Cardiovascular Laboratory, the Coronary Care Unit and the Laboratory for Experimental Surgery, University of Louvain, Belgium
› Author Affiliations
Further Information

Publication History

Publication Date:
16 February 2018 (online)

    Permissions and Reprints

Digital “on-line” QRS detection and classification in monitoring units is often performed by an algorithm using the ECG spatial velocity. Our study was undertaken to provide as well as to justify the choice of the value given to each parameter of such an algorithm (Gerlings et al., Comput. biomed. Res.5: 14—24, 1972). It was made on 5-minute ECG recordings sampled from a single lead on 17 patients. The following parameters were studied and their error function (QRS not detected, badly classified, and detected but non-existent) thoroughly investigated. For the QRS detection which was based on three parameters, para WD (window of derivative), para TD (threshold of derivative) and para TP (threshold of sampled points), the choice of WD at 24 ms was found to be critical and provided the optimal detection at 96.9%. For the QRS classification only para TolP (tolerance of sampled points)was shown to be important; when set at .5 para TolP gave the optimal classification at 97%. The method can also be used to investigate the response of the algorithm to particular QRS and to propose for such cases a specific and optimal value for each parameter.

Die digitale »on-line« QRS-Erkennung und Klassifizierung in Monitor-Einheiten erfolgt oft durch einen auf der räumlichen EKG-Vektorgeschwincligkeit basierenden Algorithmus.

Die vorliegende Studie hatte das Ziel, die Werte jedes Parameters eines solchen Algorithmus zu be-stimmen und zu rechtfertigen (Gerlings et al., Comput. biomed. Res.5: 14·—24, 1972).Bei 17 Patienten wurde eine EKG-Ableitimg während 5 Minuten aufgeschrieben. Die folgenden Parameter und ihre Fehlerhaftigkeit (nicht ermittelte QRS-Zacke, falsch klassifizierte und entdeckte, aber gar nicht existierende QRS-Zacke) wurden eingehend untersucht.

Für die QRS-Erkennung, die auf drei Parametern basierte — para WD (window of derivative), para TD (threshold of derivative) unci para TP (threshold of sampled points) — erwies sich die Wahl der»WD« bei 24 m Sek. als lu’itisch; sie lieferte eine optimale Erkennung bei 96,9%. Für die Klassifizierungcler »QRS«-Zacken erwies sich nur die para TolP (tolerance of sampled points) als bedeutsam; bei einem Wert von 0,5 ergab para TolP eine optimale Klassifizierung von 97%.

Die Methode kann auch verwendet werden, um die Antwort des Algorithmus auf spezielle QRS-Zackenzu untersuchen und für solche Fälle spezifische und optimale Werte für jeden Parameter vorzuschlagen.

Keywords

ECG On-line Monitoring - Digital Processing - Spatial Velocity - Optimization

Schlüsselwörter

EKG-On line-Überwachung - Digitale Verarbeitung - räumliche Vektorgeschwindigkeit - Optimierung

*) Paper presented at the 1st Meeting of the Belgian Society for Medical Informatics, Brussels, Dec. 7, 1974.


 

  • References

  • 1 Feldman C. L, Amazeen P. G, Klein M. D, Lown B. Computer detection of ventricular ectopic beats. Comput. biomed. Res 3 1071; 666-674.
    PubMedGoogle Scholar
  • 2 Geddes J. S, Warner H. R. A PVC detection program. Comput. biomed. Res 4 1971; 493-508.
    CrossrefPubMedGoogle Scholar
  • 3 Gerlings E. D, Bowers D. L, Rol G. A. Detection of abnormal ventricular activation in a coronary care unit. Comput. biomed. Res 5 1972; 14-24.
    CrossrefPubMedGoogle Scholar
  • 4 Haywood L. J, Murthy V. K, Harvey G. A, Saltzberg S. On-line real time computer algorithm for monitoring the ECG waveform. Comput. biomed. Res 3 1970; 15-25.
    CrossrefPubMedGoogle Scholar
  • 5 Pryor T. A, Ridges J. D. A computer program for stress test data processing. Comput. biomed. Res 7 1974; 300-368.
    PubMedGoogle Scholar
  • 6 Rosenbrocik H. H. An automatic method for finding the greatest or least value of a function. Comput. J 3 1960; 175-184.
    CrossrefPubMedGoogle Scholar
  • 7 Wartak J, Milliken J. A, Lywood D. W. Theoretical and practical aspects of analog-to-digital conversion of the electrocardiogram. Medical Res. Eng 9 1970; 21-23.
    PubMedGoogle Scholar
  • 8 Wartak J, Milliken J. A, Karchmar J. Computer program for pattern recognition of electiocardiograms. Comput. biomed. lies 4 1970; 344-374.
    PubMedGoogle Scholar