Removal of CPR-related artefacts in VF-ECG

The present research proposal has the aim (A) to develop an easily accessible database of emergency medical ECG data, (B) to develop mathematical algorithms, which allow to remove CPR-related artifacts from the ventricular fibrillation ECG and (C) to provide a cantilever-based force measurement to record the force applied on the sternum during reanimation conditions. Presently, we rely on about 75 emergency medical VF-ECG datasets of good quality (375 Hz, 12 bit), 200 emergency medical VF-ECG datasets of standard quality (100 Hz, 8 bit) and about 200 animal VF-experiments (1000 Hz, 12 bit). The CPR-artifact removal algorithms, the corresponding software and the force-measurement device, which will be developed should be incorporated into future versions of defibrillators, together with algorithms, which predict defibrillation success based on the VF-ECG signal ("fibrillation scoring"). Tests using simple bandpass filters have not been successful in performing CPR-artifact removal. More sophisticated methods are necessary, using the tonal structure of CPR (around 80 cardiac massages/min = 1.33 Hz and its overtones) and the pulse-like form of these artifacts in time domain and/or additional information based on the contact pressure of the manual reanimation or the blood pressure of the patient. Our development of CPR-removal software is an innovation, the application and use of which in defibrillators would increase the possibilities of extracting information from the VF-ECG. The innovation potential of such combined software, performing CPR-removal and fibrillation scoring, will be considerable. Such combined software will permit to continue cardiac massage during the defibrillator analysis phase, reducing the potentially lethal hands-off times (see e.g. Steen S et al. "The critical importance of minimal delay between chest compressions and subsequent defibrillation", Resuscitation 2003). Our research group: (A) collects and documents preclinical data in a joint study with defibrillators from WelchAllyn, (B) has the mathematical know-how for development of CPR-removal algorithms, (C) has developed and improves fibrillation scoring algorithms, (D) has the know-how for the development of appropriate databases, (E) and has the means to test new soft- and hardware in a preclinical setting. Our research is at the interface between fundamental and applied research, combining medical knowledge with mathematical and software know- how.

Ischemic cardiac disease is one of the leading causes for death in the Western world. In Europe, cardiovascular disease accounts for 40% of all cases of death of persons younger than 75 years. For adults suffering from cardiovascular disease 60% of deaths are due to sudden cardiac death. The yearly incidence of out-of-hospital sudden cardiac arrest is about 38/100000 persons. Approximately 17/100000 persons display ventricular fibrillation (VF) at the arrival of emergency medical services. Therefore ventricular fibrillation is a particular challenge for the health system. Early defibrillation, optimally performed after cardiopulmonary resuscitation (CPR), is considered to be the only effective treatment for VF and is one of the keys step in the "chain of survival" in cardiac arrest. A critical aspect of the chain of survival is the cumulative no-flow time (NFT), during which CPR is interrupted, e.g., for taking emergency action other than CPR. NFT is, for example, due to checking the ECG rhythm of a patient before a defibrillation shock. According to the 2010 CPR-guidelines, NFTs should be minimized, because by impairing coronary blood perfusion, they reduce the survival rate. NFT might be substantially reduced in case automated external defibrillators (AED) would analyze the ECG rhythm during ongoing CPR. when motion artefacts disturb the ECG signals. Sophisticated signal processing techniques may be used to filter such artifacts from the ECG during ongoing CPR. In part, the implementation of such algorithms into a defibrillator has been realized in the Zoll R Series device. In the present project "CPR-filtering algorithms for ECG-analysis" we used different mathematical techniques to remove CPR-related artefacts. Time-frequency methods, Kalman filtering, independent component analysis and Gabor analysis were the main techniques used. In addition to the development of CPR-artefact removal algorithms, using a unique pig model we generated an ECG-database containing emergency ECGs corrupted with different real artefacts. Also a MATLAB-type toolbox was implemented. The results were published in leading journals, the most important publications being as follows: Rheinberger at al., IEEE Transactions on Biomedical Engineering 55: 130-137 (2008) Werther et al., IEEE Transactions on Biomedical Engineering 65, 320 - 327 (2009) Werther et al., Resuscitation 80, 1301 - 1307 (2009) Amann et al., BioMedical Engineering OnLine 9, 2 (2010) Granegger et al., Resuscitation 81, 730 - 736 (2010) Xia et al., IET Signal Processing 4, 650 - 657 (2010) Granegger et al., Resuscitation 82, 79 - 84 (2010).