Abstract 68: Compression Depth Was Correlated with SNR in VF but Not in Nonshockable Rhythms During CPR
Introduction: Artifacts produced by chest compression (CC) during cardiopulmonary resuscitation (CPR) preclude reliable automated ECG rhythm analysis. Several filtering techniques have been proposed to remove the artifact with the use of compression related signals as reference. In order to test the performance of the filters, CPR-corrupted signals were usually generated at different signal-to-noise ratio (SNR) levels and the improvements in SNR were evaluated. However, SNR in real cardiac arrest data and the relationship between compression depth (CD) and SNR are still unknown.
Method: ECG, together with CD signals were recorded through defibrillators in 146 patients who experienced cardiac arrest and CPR. A total of 306 segments (6 seconds length, sample rate 250Hz), including 152 ventricular fibrillation (VF), 91 pulseless electrical activity (PEA) and 63 asystole were analyzed. Each segment consisted of 3 seconds corrupted and adjacent 3 seconds artifact-free signal. The power of signal was calculated from the artifact-free signal and the power of artifact was obtained through subtracting the power of corrupted ECG by the power of artifact-free signal. The relationship between CD and SNR was tested with Pearson correlation coefficients.
Results: The average CD was 4.76±1.24cm. SNR of asystole was significantly lower compared to that of VF and PEA (-21.93±7.37 dB vs. -5.04±8.53 dB and -3.70±7.16 dB, p<0.01). As shown in the Figure, the correlation coefficient was -0.16 (p<0.01) between CD and SNR. When each of the rhythms was investigated individually, negative correlation between CD and SNR was only observed in VF (r =–0.21, p < 0.01).
Conclusion: In this patient population, SNR of corrupted ECG was not correlated with CD when the underlying rhythms were non-shockable. This may, at least in part, accounted for the reported high sensitivity but low specificity when adaptive filters were used to remove artifact using CD related signals as reference.
Author Disclosures: Y. Li: None. Y. Gong: None. B. Chen: None. M. He: None.
- © 2014 by American Heart Association, Inc.