Difracci´ on de Rayos X (XRD)

The original contributions of the present thesis can be summarized as follows.

a) The creation of a dynamic 2.5D bioimpedance model dedicated to investigating the influence of cardioballistic effects (heart motion) on the EIT heart signal. b) The demonstration via simulations on the model of the feasibility of tracking

changes in CO by means of EIT with good accuracy – despite the dominantly cardioballistic origin of the EIT heart signal – and the identification of factors that may affect this accuracy.

9.5. Limitations and future work c) The creation of a 4D bioimpedance model of the pulmonary vasculature capable of realistically simulating pulmonary hemodynamics in various cases of pulmonary hypertension.

d) The development of a novel non-invasive and unsupervised PAP monitoring method based on the use of EIT and the measurement of changes in pulmonary pulse wave velocity.

e) The demonstration, via simulations on the aforementioned model, of the feasibility of tracking changes in PAP by means of the proposed method in various cases of pulmonary hypertension.

f) The validation in vivo of the proposed method in healthy subjects undergoing acute changes in PAP through hypoxia.

9.5

Limitations and future work

9.5.1 EIT-based CO monitoring

Our investigations have provided novel insights about the genesis of the EIT heart signal, and helped to determine the physiological and metrological requirements for an accurate measurement of CO by EIT. However, our observations remain limited by the absence of experimental validation. In our opinion, future work should therefore mainly focus on carrying out a clinical trial, e.g. before and after anesthesia in patients undergoing cardiac surgery, using PAC thermodilution CO measurements as reference. Patients with valvular insufficiencies or large pericardial effusion should be excluded. Changes in SV could be induced using the passive leg raising maneuver, used to predict fluid responsiveness in critically ill patients [116]. This maneuver consists in elevating the legs of the patient from the supine position by a 30–60◦angle, which allows increasing the cardiac preload and thereby increasing SV. The so-called semi-recumbent position has the opposite effect: the patient’s trunk is raised from the supine position by a 30–60◦ angle, thus decreasing the cardiac preload and thereby decreasing SV. Using these two maneuvers along with the baseline supine position would therefore allow generating SV variations to evaluate the trending ability of EIT. These maneuvers are non-invasive and easy to perform, and could therefore be repeated twice to increase the number of paired EIT-PAC measurements. Furthermore, they could also be repeated after anesthesia, as the baseline CO is likely to have changed. As a results, a series of 6 to 12 paired EIT-PAC measurements would be obtained for each patient, which is sufficient to assess trending ability [36].

According to our observations, careful consideration should be given to the following aspects regarding the EIT measurements:

Chapter 9. Synthesis

a) The EIT belt should be placed in the oblique plane as described in [171] to maximize ventricular contribution to the EIT heart signal.

b) If feasible during the experimental protocol, typically before anesthesia at baseline condition, the belt should initially be placed∼1.5 cm below the desired position, then ∼1.5 cm above, and finally at the desired position. At each position, EIT measurements paired with PAC measurements should be performed. These mea- surements should allow quantifying the influence of belt displacements on SV estimation experimentally.

c) During the remainder of the protocol, particular attention should be paid to mini- mize the displacements of the belt.

d) During data analysis, we suggest the use of respiration-gating. The influence of respiratory activity on the accuracy of the EIT-based measurement of CO should be formally quantified by comparing the estimation errors obtained when respiration- gating is or not used.

9.5.2 EIT-based PAP monitoring

A novel method for the non-invasive monitoring of PAP by EIT has been proposed and validated both through simulations and experimentally. However, our experimental study comprises some limitations, such as the use of Doppler echocardiography as reference instead of the gold standard PAC. Although validated against PAC measure- ments [10, 37, 108], echocardiographic measurements of the PAP can still be inaccurate in the individual patient [56]. A future clinical study in patients should therefore use the PAC as reference.

In such a study, PAP variations could for instance be induced after anesthesia by de- creasing the inspired oxygen fraction in the ventilator, causing hypoxic pulmonary vasoconstriction and increases in PAP. Alternatively, decreases in PAP could be induced via inhaled nitric oxide, a selective pulmonary vasodilator used for the treatment of patients with pulmonary hypertension and hypoxemia [76]. As the PAC and EIT both provide continuous beat-to-beat readings, a large amount of paired PAC-EIT measure- ments could thus be obtained and allow testing the beat-to-beat trending ability of EIT for PAP monitoring.

Another particularly important aspect to be investigated is the stability over time of the PAT-to-PAP calibration, in particular for the long-term monitoring of patients with chronic pulmonary hypertension. This implies testing:

a) The influence of a different positioning of the EIT belt, which directly affects the calibration itself as it affects the PAT;

9.6. Conclusion