CPR Artifact Reduction in the Human ECG by Using Constrained Independent Component Analysis

State-of-the-art automated external defibrillators (AEDs) require interruption of cardiopulmonary resuscitation (CPR) to analyze the ECG. In order to increase the defibrillation success rate, interruptions of chest compressions should be avoided. Various algorithms are available for reducing CPR artifacts. We recently could show that standard instantaneous ICA algorithms are at least as successful in removing these artifacts from the ECG as second channel methods. Constrained independent component analysis (cICA) is an ICA algorithm which incorporates prior information about the signal, like a second channel. However, to our knowledge, artifact removal from the CPR corrupted ECG has not been tried using cICA. As a new approach to solve the problem of removing CPR induced noise, the use of cICA is evaluated in this paper. Using the corrupted signals as obtained in a porcine model, we tested a cICA algorithm. After applying cICA to corrupted signals with very low SNR (<-10dB), sensitivity increased from 72% (corrupted signal) to 94% and specificity from 42% to approx. 82%, taking the AEDs decision whether the rhythm is shockable or not. When checking the similarity between the original, the corrupted and the reconstructed signal, the computed correlation values indicated a statistically significant improvement compared to the corrupted signal. We conclude that cICA should be useful in automatically separating the artifacts from the corrupted ECG.