On the Feasibility of Automated Mechanical Ventilation Control Through EIT

Objective: This paper aims to demonstrate the feasibility of coupling electrical impedance tomography (EIT) with models of lung function in order to recover parameters and inform mechanical ventilation control. Methods: A compartmental ordinary differential equation model of lung function is coupled to simulations of EIT, assuming accurate modeling and movement tracking, to generate time series values of bulk conductivity. These values are differentiated and normalized against the total air volume flux to recover regional volumes and flows. These ventilation distributions are used to recover regional resistance and elastance properties of the lung. Linear control theory is used to demonstrate how these parameters may be used to generate a patient-specific pressure mode control. Results: Ventilation distributions are shown to be recoverable, with Euclidean norm errors in air flow below 9% and volume below 3%. The parameters are also shown to be recoverable, although errors are higher for resistance values than elastance. The control constructed is shown to have minimal $H^1$ seminorm resulting in bounded magnitudes and minimal gradients. Conclusion: The recovery of regional ventilation distributions and lung parameters is feasible with the use of EIT. These parameters may then be used in model based control schemes to provide patient-specific care. Significance: For pulmonary-intensive-care patients mechanical ventilation is a life saving intervention, requiring careful calibration of pressure settings. Both magnitudes and gradients of pressure can contribute to ventilator induced lung injury. Retrieving regional lung parameters allows the design of patient-specific ventilator controls to reduce injury.