Noninvasive Inference of Airway Network Geometry From Broadband Lung Reflection Data

Abstract

An inference of the summed cross-sectional area of the airway network of the lung as a function of distance from the airway opening can be obtained noninvasively by inversion of high-frequency reflection response data with the Ware-Aki algorithm. This method is critically evaluated. The response of branching models of the complete tracheo-bronchial tree are simulated and inverted to produce an area versus distance inference. The models incorporate branching asymmetry, nonrigid walls, and a viscous gas. Direct comparison of the area inference with the anatomic area of the model indicates that the inference is markedly influenced by nonrigidity of the airway walls and branching asymmetry, and that the inference does not possess a simple anatomic correlate. Empirical uses of the inference method are not precluded.