• Tobias Werther, Andreas Klotz, Günther Kracher, Michael Baubin, Hans G Feichtinger, Hermann Gilly, and Anton Amann.
• Faculty of Mathematics, University of Vienna, A-1090 Vienna, Austria. tobias.werther@univie.ac.at
• IEEE Trans Biomed Eng. 2009 Feb 1;56(2):320-7.

Background And ObjectiveWe present an algorithm for discarding cardiopulmonary resuscitation (CPR) components from ventricular fibrillation ECG (VF ECG) signals and establish a method for comparing CPR attenuation on a common dataset. Removing motion artifacts in ECG allows for uninterrupted rhythm analysis and reduces "hands-off" time during resuscitation.Methods And ResultsThe current approach assumes a multichannel setting where the information of the corrupted ECG is combined with an additional pressure signal in order to estimate the motion artifacts. The underlying algorithm relies on a localized time-frequency transformation, the Gabor transform, that reveals the perturbation components, which, in turn, can be attenuated. The performance of the method is evaluated on a small set of test signals in the form of error analysis and compared to two well-established CPR removal algorithms that use an adaptive filtering system and a state-space model, respectively.ConclusionWe primarily point out the potential of the algorithm for successful artifact removal; however, on account of the limited set of human VF and animal asystole CPR signals, we refrain from a statistical analysis of the efficiency of CPR attenuation. The results encourage further investigations in both the theoretical and the clinical setup.

22 keywords...

hide…