A new method to detect ventricular fibrillation from CPR artifact-corrupted ECG based on the ECG alone

During cardiopulmonary resuscitation, chest compressions (CCs) introduce mechanical activity in the ECG and thus preclude a reliable electrocardiographic (ECG) rhythm diagnosis. To achieve a reliable rhythm analysis, chest compression must therefore be interrupted, and therefore, the probability of the restoration of spontaneous circulation (ROSC) is adversely affected. In recent years, a number of algorithms have been developed to distinguish ventricular fibrillation (VF) rhythm from normal sinus rhythm (SR) without chest compression (CC) interruptions. However, the implementation of most of these algorithms relies on the acquisition of reference signals that are strongly correlated with CC artifacts and makes additional hardware alteration inevitable. In the present work, a novel method (the enhanced LMS method) that effectively suppresses CPR artifacts and can easily use the corrupted ECG signal alone is developed for the reliable detection of the VF rhythm during uninterrupted CCs. The enhanced LMS method was tested using mixtures of CC artifacts and real out-of-hospital ECG recordings for different corruption levels, and it was compared with other established algorithms that use the corrupted ECG signal alone, including the morphology consistency evaluation algorithm and the adaptive stop-band filtering algorithm. The validation results indicate that the enhanced LMS method has superior performance in VF/SR rhythm classification under different artifact interference levels. It is shown that the VF rhythm can be reliably detected using only the corrupted ECG alone. The novel method proposed in this study is promising for identification VF from SR with no hardware alterations for clinical cardiopulmonary resuscitation practice.