Research On Automatic Monitoring Algorithm For Transthoracic Impedance Signal In Cardiopulmonary Resuscitation

Cardiopulmonary resuscitation(CPR) is the only effective way to rescue patients with cardiac arrest. Chest compressions(CCs) are important means for the restoration of spontaneous circulation during CPR. CC detection and real-time feedback can provide rescuers with reference for CPR and help perform CPR correctly. Previous research has shown that the compression rate and compression depth during CC can reflect in the transthoracic impedance(TTI) signal and instantaneous compression frequency can be accurately estimated. The guidance of CCs by detecting and analyzing of TTI signal is of great importance for ensuring CPR quality.According to the characteristics of TTI signal waveform during CPR, this paper proposed two kinds of automatic detection algorithm to achieve the ideal classification of TTI signal and quality evaluation of CPR through various research and experiment. This paper made research on the following aspects:①Aiming at the characteristics of the TTI signal, a filtering method of TTI signal based on wavelet threshold and morphology is designed combined with advantages and disadvantages of conventional signal preprocessing methods. Firstly, opening and closing operations is conducted and average add follows to obtain baseline drift signal, baseline drift signal is removed from the original signal, finally, through the wavelet soft threshold filter, the TTI signal filtering the high frequency noise and baseline drift is obtained.②Due to the disadvantage of existing locating algorithm for TTI signal waveform, this paper proposed a search algorithm based on a multiresolution window for waveform marking. The algorithm adopted different resolutions for processing the same signal, and the local optimal values under different resolutions could be determined and then combined to obtain more precise optimal values.③This paper proposed an automatic detection algorithm for TTI signal based on pattern recognition. The TTI signals are reprocessed by denoising algorithm based on wavelet and morphology firstly. Then the potential compression and ventilation waveforms are located using the searching algorithm of multiresolution window, and extraction weighted height Hn and width L as the features. Finally, linear discriminant analysis algorithm is used to classify and recognize the located compression and ventilation waveforms. The results show that the algorithm achieved a good classification results, and the calculation is simple, processing speed is fast.④This paper proposed automatic detection algorithm for TTI signal based on k-means clustering algorithm based on density weighting and preference information. After the TTI signals are preprocessed by de-noising algorithm based on wavelet and morphology,the potential compression and ventilation waveforms were marked by using the searching algorithm of multiresolution window. Then the difference between the adjacent waveforms and the ratio of the power of each section to the amplitude of the original one were taken as specificity and robustness features besides the height H and width L. Finally, k-means clustering algorithm based on density weighting and preference information was used to recognize and classify the compression and ventilation of the marked waveforms. The results showed that this algorithm reached high recognition accuracy and sensitivity, had good robustness and met the need for real-time detection.This paper provided a novel theory and method for the real-time monitoring of TTI signal during CPR, laid the foundation for evaluating the CC quality and monitoring in out-of-hospital CPR.