Cardiorespiratory couplings: Investigating the respiration influence onto the ECG

Description

It is a well-known fact that respiration significantly influences the electrocardiogram (ECG). Since in modern ECG signal analysis even small and subtle morphological changes are important for decision making, this influence should not be neglected, otherwise risking a misinterpretation of the results. Therefore, in this work, we investigate the coupling between the two biomedical signals, specifically the respiration-induced morphological changes of the ECG. For that reason the ECG was represented using so-called adaptive Hermite functions, allowing to describe the shape of single ECG beats with just a few coefficients. In order to determine and subsequently eliminate the influence of the respiration onto the ECGs' morphology, a linear model was identified, using the respiration signal as input and the above-described coefficients as output signal. The method was verified with artificially generated data and subsequently tested for four subjects on real ECG data. The results show a clear coupling between the respiration and the coefficients describing the ECG shape. However, as expected, the coupling strongly depends on the subject, not allowing to define a general model between the two biomedical signals. After minimizing the respiration influence onto the ECG, the shapes of the original and the reconstructed (clean) ECGs clearly differ from each other, which emphasizes the importance of considering the respiration for ECG morphology analysis. Summarizing, this work provides a first attempt for decoupling the respiration from the shape of the ECG using a rather simple linear model. It is not known yet, whether this model is best suited and therefore more complex, non-linear models should be tested as well. Furthermore, the experiments were only carried out for controlled breathing and consequently must be investigated for spontaneous breathing. Additionally, this work initiates investigations on ECG shape couplings to other biomedical signals.

Period	29 Nov 2018
Event title	4. Kepler Science Day
Event type	Conference
Location	AustriaShow on map