Measurement Principles for a real-time Cardiac Volume Sensor

Abstract

Patients suffering from heart failure cannot provide sufficient cardiac output for organ perfusion, such that effective medical treatment is required. Three therapies for the treatment of heart failure have been clinically established: pharmacological treatment, heart transplantation, and implantation of a left-ventricular assist device (LVAD). Despite the efficacy of these therapies, survival rates remain unsatisfactorily low. The continuous monitoring of hemodynamic parameters can improve survival, as it supports timely and effective clinical decision making. The hemodynamics of the healthy heart are very sensitive to the left-ventricular (LV) volume. Hence, the LV volume is a promising hemodynamic parameter for clinical decision making. In addition, an LV volume measurement could be used for physiological feedback control of an LVAD, presumably increasing his or her quality of life and reducing the probability of adverse events. The aim of this thesis was to identify key requirements for an LV volume sensor, investigate possible measurement principles and evaluate the three most promising sensor concepts in terms of their sensitivity and accuracy. The LV volume cannot easily be measured remotely, because the requirements for a real-time portable sensor are highly complex. An LV volume sensor needs to be implemented in such a way that traumatic injury is avoided and the areas of foreign surfaces in the body are not increased. The sensor should be small enough to be safely implanted or attached to the body surface. The LV volume measurement should be continuous and robust to changes in heart geometry, posture or hematocrit. Computer tomography, magnetic resonance imaging, various forms of impedance measurement, pressure measurement, echocardiography and strain sensors have been proposed to estimate the LV volume. However, none of them meets all of the above requirements. An implantable, biocompatible and robust LV volume sensor remains to be developed. Three concepts for the real-time LV volume measurement are proposed in this thesis: Acoustic resonance, ultrasonic distance and the QRS amplitude of the electrocardiogram. The concepts were evaluated in a testing environment suitable to their current stage of development: in-silico, in-vitro, in-vivo and in the setting of an experimental clinical study. All measurement priniciples were sensitive to the LV volume. The achievable LV volume accuracies were assessed using a Bland-Altman analysis. The accuracies for the acoustic concept could not be evaluated as the principle was only assessed in-silico and lacked the possibility to account for noise. The ultrasonic distance approach yielded estimation accuracies for the LV volume smaller than 20% in human heart phantoms in vitro. The QRS-amplitude in vivo measurement rendered LV volume estimation accuracies smaller than 20%. The experimental clinical study revealed a small, but significant correlation between the QRS amplitude on the body surface and the mean pulmonary arterial pressure. The three concepts presented are all atraumatic, small enough, capable of real-time measurement and should be sufficiently robust as they can be placed close to the heart. The QRS amplitude is the most promising concept to be implemented in the near future, particularly because of the electrode size. The ultrasonic distance measurement is equally convincing in terms of accuracy, but requires more effort for miniaturization and efficient data processing. The influence of hematocrit changes on both measurement principles should be investigated carefully in subsequent studies. In conclusion, continuous measurement of LV volume is possible and will likely increase survival rates in heart failure patients in the future.