Non-invasive hemodynamic monitoring by electrical impedance tomography

The monitoring of central hemodynamic parameters such as cardiac output (CO) and pulmonary artery pressure (PAP) is of paramount clinical importance to assess the health status of the cardiovascular system. However, their measurement requires the insertion of a pulmonary artery catheter, a highly invasive procedure associated with non-negligible morbidity and mortality rates. In this thesis, we investigated the clinical potential of electrical impedance tomography (EIT) - a radiation-free medical imaging technique - as a non-invasive alternative for the measurement of CO and PAP. In a first phase, we investigated the potential of EIT for the measurement of CO. This measurement is implicitly based on the hypothesis that the EIT heart signal (the ventricular component of the EIT signals) is induced by ventricular blood volume changes. This hypothesis has never been formally investigated, and the exact origins of the EIT heart signal remain subject to interpretation. Therefore, using a model, we investigated the genesis of this signal by identifying its various sources and their respective contributions. The results revealed that the EIT heart signal is dominated by cardioballistic effects (heart motion). However, although of prominently cardioballistic origin, the amplitude of the signal has shown to be strongly correlated to stroke volume (r = 0.996, p < 0.001; error of 0.57 +/- 2.19 mL). We explained these observations by the quasi-incompressibility of myocardial tissue and blood. We further identified several factors and conditions susceptible to affect the accuracy of the measurement. Finally, we investigated the influence of the EIT sensor belt position on the measured heart signal. We observed that small belt displacements - likely to occur in clinical settings during patient handling - can induce errors of up to 30 mL on stroke volume estimation. In a second phase, we investigated the feasibility of a novel method for the non-invasive measurement of PAP by EIT. The method is based on the physiological relation linking the PAP to the velocity of propagation of the pressure waves in the pulmonary arteries. We hypothesized that the variations of this velocity, and therefore of the PAP, could be measured by EIT. In a bioimpedance model of the human thorax, we demonstrated the feasibility of our method in various types of pulmonary hypertensive disorders. Our EIT-derived parameter has shown to be particularly well-suited for predicting early changes in pulmonary hemodynamics due to its physiological link with arterial compliance. Finally, we validated experimentally our method in 14 subjects undergoing hypoxia-induced PAP changes. Significant correlation coefficients (range: [0.70, 0.98], average: 0.89) and small standard errors of the estimate (range: [0.9, 6.3] mmHg, average: 2.4 mmHg) were found between our EIT-derived systolic PAP and reference systolic PAP values obtained by Doppler echocardiography. In conclusion, there is a promising out