Developing integrative computational models of pulmonary structure.

Abstract

Integrative computational modeling of the pulmonary system aims to incorporate interactions between the lung's subsystems by means of a hierarchy of structural and functional models. This requires detailed imaging-based data, along with a wide range of functional information from experiments. Advances in computed tomography imaging technology ensure that high-resolution data are now readily available upon which the structure of these models can be based. We present methods for constructing anatomically realistic finite element models of interrelated pulmonary structures from such data. Segmented human lung lobe data are fit to high-order (cubic Hermite) volume elements. Meshes for the conducting airways and pulmonary arteries and veins are constructed within the lobe mesh, using a combination of fitting to imaging data and a bifurcating-distributive algorithm. The algorithm generates an airway-consistent mesh within a host volume, and this airway mesh is then used as a template for generating blood vessel models. The lung parenchyma is modeled as a space-filling three-dimensional (3D) Voronoi mesh, with generated geometry consistent with the alveolated airway structure. Pulmonary capillaries are generated over the alveolar model, as a 2D Voronoi mesh. These structural models have been compared extensively with morphometric data to verify that their geometry is representative of the pulmonary system. The models are designed to be integrative: they relate multiple structural systems within the same individual, and their use as computational meshes allows application of spatially distributed properties.