Branching airway network models for analyzing high-frequency lung input impedance
The input impedance of the lung (Zin) at high frequencies (> 100 Hz) is a complex function of the airway geometry and the mechanical properties of the airway walls. To investigate how the purely geometrical factors influence Zin, we measured Zin between 16 and 1,520 Hz in six dried dog lungs with the forced oscillation technique. In each of the lungs we found three resonances, at 36 +/- 5, 648 +/- 100, and 1,289 +/- 150 Hz, and at least two antiresonances (relative maxima in the real part of Zin), at 372 +/- 60 and 1,105 +/- 110 Hz. These data were fit with models featuring a detailed asymmetric branching network of the airways obtained from morphometric data published by Horsfield et al. (J. Appl. Physiol. 52: 21–26, 1982). On the basis of low-frequency (< 100 Hz) data alone, we first established a model of the acini, which was then attached to the end of the airway branching model. With a single scaling factor for the radius and length of the airways, the fit was unsatisfactory. Using sensitivity analysis techniques we determined which candidate variables of the structural model could influence Zin in a manner to improve the fit. We found that a two-parameter model accounting for separate central and peripheral airway diameter scaling provided a reasonable fit to Zin. On average the model required central diameter scaling close to unity (0.94 +/- 0.09), and the peripheral diameter scaling factor was 0.87 +/- 0.38. Over a range of parameter values that we believed were physiologically reasonable (i.e., scaling factors between 0.5 and 1.5), a single set of parameter values was found in all lungs. These results suggest that structurally based inverse models of Zin that include multiple antiresonances may provide information about airway geometry.