Systematic investigation of event detection, hardware, and models for impedance cardiography

Date of Award




Degree Name

Doctor of Philosophy (Ph.D.)


Biomedical Engineering

First Committee Member

Joachim H. Nagel, Committee Chair


Impedance cardiography (ICG) has been acknowledged as a valuable non-invasive tool for monitoring relative stroke volume (SV) changes. However, the ICG has not been fully accepted as a diagnostic tool since the absolute value of SV obtained by this technique is still questionable in particular patient groups. Through a systematic investigation, three sources of error were found to contribute to this inconsistency. They are: occasional inadequacy of the ICG amplifier, unreliable event detection techniques, and the shortcomings in the existing ICG model. The commonly used method for measuring the signal amplitude (maximum look up) was found to introduce a significant amount of error when the signal to noise ratio (SNR) is low. The commonly used event detection techniques cause substantial errors in the detection of the B point. The insufficiency of the dZ/dt circuit in the most commonly used commercially available ICG amplifier (Minnesota Impedance Cardiography) was found to cause errors in the amplitude and event latency. Consequently, a significantly smaller SV, a longer pre-ejection period, and a larger Heather index are obtained.In this dissertation, signal processing techniques, especially adaptive matched filtering event detection and correlative amplitude determination have been implemented and could be demonstrated to improve the reliability and accuracy of event detection and signal amplitude estimation. Additionally, a new method to determine the upper band-limit, which is based on the consideration of the inherent variance of physiological signals, has been developed and used in the confirmation of the assumption that the ICG signal bandwidth increases as the heart rate increases. Necessary specifications for the ICG amplifier have been established that may lead to standardization of impedance cardiography, which is pre-required for clinical applications in the future. The ICG gives a reasonable estimation of SV for some patient groups while it fails in other groups. Two assumptions in the formulation of the existing SV equations have been shown to be questionable, and this might explain the difficulties. With improved signal processing techniques, improved amplifier, and a better knowledge of the ICG signal spectrum, the shortcomings of the existing model and equations can be examined more reliably and accurately in future studies.


Engineering, Biomedical; Health Sciences, Radiology

Link to Full Text