Ventricular fibrillation detection by a regression test on the autocorrelation function

Research output: Contribution to journalArticlepeer-review


The paper investigates quantitative differences in the signal characteristics of ventricular fibrillation (VF) and other cardiac arrhythmias. The analysis procedure comprises two steps: calculation of a short-term autocorrelation function (ACF) followed by a regression test on a plot of peak magnitudes of the ACF against lag values (the ACF/lag plot). We detect VF by testing the hypothesis that the ACF/lag plot of VF does not pass a linear regression test. Analysis of 31 separate episodes (of VF and other ventricular arrhythmias), each comprising three successive segments of 1·5s each produced the following results: (1) 100 per cent sensitivity (Se), 62 per cent specificity (Sp) and 74 per cent test efficiency (TE) after analysis of the first segment; (2) 100 per cent Se, 86 per cent Sp and 90 per cent TE after the second segment; and (3) 100 per cent Se, 100 per cent Sp and 100 per cent TE after the third segment. This method quantifies the notion that VF signals are nonperiodic with a random amplitude distribution, whereas ventricular tachycardia (VT) signals are usually periodic with more uniform amplitude distributions. Accurate discrimination and identification of VF can be very important in intensive-care settings, as well as in the design of automatic cardioverters and defibrillators.

Original languageEnglish (US)
Pages (from-to)241-249
Number of pages9
JournalMedical & Biological Engineering & Computing
Issue number3
StatePublished - May 1 1987


  • Analysis
  • Autocorrelation
  • Defibrillation
  • Performance
  • Regression
  • Signal
  • Ventricular fibrillation

ASJC Scopus subject areas

  • Biomedical Engineering
  • Computer Science Applications

Fingerprint Dive into the research topics of 'Ventricular fibrillation detection by a regression test on the autocorrelation function'. Together they form a unique fingerprint.

Cite this