Low-frequent lung impedance estimation on breathing patients

The respiratory impedance Z is a widely used quantity for the monitoring of lung diseases. Z is defined as the frequency dependent ratio between pressure and flow at the mouth of the patient. To measure Z, the forced oscillation technique is used. In the forced oscillation technique, Z is determined by superimposing small amplitude (in the order of 0.1kPa) pressure oscillations on the normal breathing and measuring the resulting flow signal generated by the patient. This generated flow signal will contain both the breathing signal and the response of the respiratory impedance to the imposed pressure oscillations. A lot of useful information for disease monitoring is contained in the frequency range of spontaneous breathing (0.1-1Hz). Therefore the goal of our setup is to obtain the respiratory impedance in this frequency range while the patient can continue normal breathing. This introduces two main difficulties. First, the breathing signal is present in the flow signal and is considered as a disturbance. Therefore it needs to be eliminated without losing useful information about the respiratory impedance itself. Second, after eliminating the breathing disturbance, a quasi-periodically time varying respiratory impedance needs to be modeled. Also the rate of time-variation coincides with the frequencies of interest for the modeling of the respiratory impedance. To cope with this combination of complexities and to obtain an accurate estimate of the respiratory impedance we propose a combination of nonlinear estimation techniques and the local polynomial method.