Changes in cardiac indices from implanted defibrillator-stored electrograms due to acquisition and preprocessing conditions

A wide number of cardiac indices have been proposed to describe electrocardiograms (ECG) during Ventricular Fibrillation (VF), and they can be useful when analyzing electrograms (EGM) stored in Implantable Cardioverter Defibrillator (ICD) during spontaneous VF. However, the dependence of their measurement on acquisition and preprocessing conditions has not been so far statistically quantified. We propose a systematic procedure based on nonparametric bootstrap resampling methods to obtain standard errors and confidence intervals for a test. This test detects changes in the statistical distribution of cardiac indices from ICD-stored EGM during VF, due to discrepancies in acquisition and preprocessing conditions. As an application example, significant changes in the distributions of selected spectral indices due to lead configuration were found by comparing measurements obtained from simultaneously recorded unipolar and bipolar EGM during VF. Our nonparametric bootstrap approach can be readily applied to the measurement of cardiac indices, allowing us to study their changes under a diversity of conditions in a systematic way.