Abstract 241: Pre- and Postshock Electrogram Features in Human VF and the Occurrence of Refibrillation
Background: During cardiac resuscitation, recurrence of VF (or refibrillation) is a common problem that could affect the survival rates. Predicting the occurrence of refibrillation by analyzing the pre-shock/post-shock waveform would enable the EMS personnel to administer anti-arrhythmic drugs that could prevent refibrillation and could help in optimizing resuscitation outcomes.
Objective: To evaluate wavelet based features extracted from pre-shock waveform and QT interval measurements from post-shock waveform in predicting refibrillation.
Method: The human VF database used in this study consisted of 14 successful, 7 refibrillation, and 13 unsuccessful out-of-the-hospital VF tracings extracted from the Zoll defibrillators recorded by the Toronto area EMS personnel. For the wavelet analysis (continuous wavelet transform) we used the 10s pre-shock waveforms from all the 34 cases that were not corrupted by the CPR artifacts. Wavelet features that indirectly measure signal (i.e., electrogram) complexity were extracted as features. A two stage classification was performed, in the first stage the 34 cases were classified into two classes i.e., CLASS I: 14 succ. + 7 refib. cases and CLASS II: 13 unsuccessful cases, and in the second stage the correctly classified CLASS I was further classified into succ. and refib. categories. The QT interval analysis used the first 3 beats of the post-shock waveforms of the 7 refib. cases and verified if a decrease in QT interval was observed.
Results: For the first stage CLASS I and CLASS II categories, we obtained a classification accuracy of 76.5 % and for the second stage succ. and refib. categories, we obtained a classification accuracy of 75%. However considering only the prediction of refib., we could correctly classify 4 out of the 7 refib. cases. An automated pattern classifier was used for the pattern classification. The QT interval analysis did not yield conclusive results for this study.
Conclusions: The pre-shock electrogram features demonstrate potential in predicting refibrillation while the post-shock electrogram features were inconclusive.
- © 2011 by American Heart Association, Inc.