| Cardiac arrest(CA), also known as cardiopulmonary arrest or circulatory arrest, is a sudden stop in effective blood circulation due to the failure of the heart to contract effectively or at all. Arrested blood circulation prevents delivery of oxygen and glucose to the body. Lack of oxygen and glucose to the brain causes loss of consciousness and brain injury, which then results in abnormal or absent breathing. According to statistics, more than 300 thousand victims in USA and Canada experience Out-of-Hospital Cardiac Arrest(OHCA) each year, and it would be more than 500 thousand in China. Although numerous scientists have made a lot of efforts on improving the cardiac arrest patient survival rate, it is still disappointing that the survival rate for cardiac arrest patients is only about 10%-20% in European, 3-5% in United States, and 2% in China.Cardiopulmonary Resuscitation(CPR) is a major treatment to deal with cardiac arrest patients. CPR is an emergency procedure performed in an effort to manually preserve intact brain function until further measures are taken to restore spontaneous blood circulation and breathing in a person who is in cardiac arrest. Its main purpose is to restore partial flow of oxygenated blood to the brain and heart.Early CPR and early defibrillation are two key points in the process of resuscitation, which is critical for the restoration of spontaneous circulation(ROSC) and survival rate of patients. According to the reports, survival rate decreases 3%-4% for each minute of defibrillation delay if bystander CPR is provided. If CPR is not provided, the decrease in survival rate will reach 7%-10% per minute.With the help of automatic external defibrillator(AED), bystanders could provide efficient CPR and defibrillation; therefore reduce the delay of resuscitation for cardiac arrest patients. The AED could distinguish shockable rhythm from nonshockable rhythms automatically, and provide shock advice to rescuers.However, chest compressions are mandated to be interrupted in the current AEDs, because the artifacts produced by chest compressions reduce the accuracy of rhythm analysis. However, the duration of these compression interruptions adversely affects the ROSC. Therefore, the 2015 American Heart Association(AHA) Guidelines emphasize the minimization of the interruptions between the compression and the shock.In order to minimize the pre-shock/post-shock hands-off intervals, suppressing the chest compression induced artifacts using digital signal processing techniques is considered a promising method. High pass filter was used firstly to suppress the CPR artifact. Although the result in experimental settings was promising; the artifacts cannot be efficiently suppressed for practical application in human victims. Subsequently, numerous algorithms have been developed to suppress the chest compression related artifacts, without or with additional reference signals, such as independent component analysis(ICA), coherent line removal algorithm, Gabor multipliers, Kalman filter, adaptive filter, and multichannel recursive adaptive matching pursuit(MC-RAMP) filter. These methods have significantly improved the signal to noise ratio(SNR) of the artifact corrupted electrocardiograph(ECG) signals, as well as the accuracy of rhythm classification. However, they were still inadequate for clinical application due to the filtering residuals.1. The underlying nonshockable rhythms would be covered when severe CPR artifacts are induced. Although CPR artifact could be suppressed after filtering with certain algorithm; residual artifact component might still remain. These residuals usually act as a disorganized signal due to the filtering process, and might be easily detected as ventricular fibrillation(VF) by the rhythm analysis algorithm.2. Spiky artifacts would be induced through the displacement of defibrillation pads. The morphology of spiky artifacts is similar with QRS complexes, which cannot be reflected in the recorded reference signal such as compression depth and transthoracic impedance. Therefore, these artifacts could not be suppressed efficiently by traditional filtering algorithms. As a result, VF would be misclassified as nonshockable rhythm when the chest compression induced spiky artifacts are interpreted as QRS complexes.The aim of this study is to conduct further research for artifact suppression during uninterrupted CPR. An enhanced adaptive filtering method is proposed to suppress the CPR artifact and improve the outcome of rhythm analysis algorithm. The major contributions of this study are presented in the following points.1. An enhanced adaptive filtering method was proposed. In this method, we generated multichannel reference matrix using the recorded reference signal, and selected the optimized reference signal from the matrix for adaptive filter based on the correlation coefficients between ECG signal and reference signals. Meanwhile, a new parameter of artifact proportion, pro, was used to monitor the artifact level in ECG signal, adjust the step size of adaptive filter, and control the iterative filtering of the proposed method. Besides, a method for spiky artifact removing was proposed to suppress the spiky artifacts. Spiky artifacts were detected through the comparison between ECG signal and reference signal, and were replaced by cubic spline interpolation.2. The feasibility of SNR estimation method was investigated. A total of 10 segments asystole(ASY) recorded from OHCA patients during CPR was selected and used as pure artifact to combined with 21 segments artifact free VF and 21 segments artifact free Pulseless electrical activity(PEA) based on preset SNR ranging from-20 d B to 15 d B. The SNR of these combined ECG signals were estimated and compared with the preset SNR. The results showed no significant difference between the estimated SNR and the preset SNR.3. The relationship between pro and SNR was researched. The combined signals were used to estimate pro and compared with the preset SNR. The results showed no significant difference between pro and the preset SNR, which demonstrated that this parameter could be used to reflect the artifact level of the corrupted ECG signals.4. An experiment trial was conducted to investigate the effect of difference recorded reference signal for the proposed method. A total of 283 shockable ECG segments and 280 nonshockable ECG segments were extracted from 22 recordings in an experimental trial. After filtering with the proposed method using compression depth as the reference signal, a sensitivity of 93.3% and a specificity of 96.0% were achieved, and no significant difference were observed between the classification result using compression depth as the reference signal and transthoracic impedance.5. The performance of the proposed filtering method was evaluated by clinical data. A total of 183 shockable and 453 non-shockable segments of ECG signal, together with CPR-related reference signal, were extracted from 233 out of hospital cardiac arrest patients. The method was optimized on a training set that comprised of 85 shockable and 211 non-shockable segments, and evaluated on a testing set with 98 shockable and 242 non-shockable segments. Compared with artifact corrupted ECG signals, the SNR increased from-9.8±12.5d B to 6.2±13.3d B, and the accuracy for detecting shockable rhythms was improved from 74.1% to 92.3% after filtering with the proposed method.These results indicated that the artifact proportion pro could be used to reflect the artifact level of the corrupted ECG signals, so that the proposed method could effectively suppress the chest compression related artifacts and improve the accuracy of rhythm analysis during uninterrupted CPR,regardless of the source of reference signals. |
PDF Full Text Request