Analysis of the effect of the respiration on features extracted from the electrocardiogram and the heart sounds

Respiration is known to influence cardiac signals through the modulation of both cardiac activity and thoracic motion. The respiratory contribution on the modulation of the electrocardiogram (ECG) both in amplitude and frequency is well established. On the other side, similar respiratory-related variations may also affect heart sounds, visible in the phonocardiogram (PCG). In this study, both ECG and PCG signals were analyzed to evaluate the influence of respiration on these signals and, in particular, its impact on the timing of heart valve closures. The latter have a clinical application in noninvasively assessing the hemodynamics of the heart. In this context, the aim of this thesis is to assess whether accounting for respiratory information could improve the accuracy of Cardiac Time Intervals (CTIs) estimation. To this end, a multi-sensor device comprising ECG, PCG, and a triaxial accelerometer was used for respiratory detection. The respiratory signal was extracted from the accelerometer data, and points of maximum and minimum thoracic expansion were identified to assess the influence of respiration on cardiac latency estimation. The ECG was used as a temporal reference for cardiac latency estimation, whereas the PCG was employed to assess the respiratory impact on the timing of valve closure events. In the analyzed group of subjects, no relevant differences in the estimated valvular time of closure in different respiration phase were observed, although statistically significant conclusions could not be drawn due to the limited number of subjects. From field application perspective, these findings suggest that explicit knowledge of the respiratory signal is probably not necessary for accurate estimation of CTIs. To obtain a statistically sound answer to the main question, a more extensive study should be needed, that is beyond the scope of this thesis. Results herein reported should be considered as a pilot study that produced some of the data necessary to estimate the minimum size of the sample population to be considered for a statistically sound study.