Geometry and aerosol dynamics in a human virtual lung for the study of tuberculosis

Despite Tuberculosis (TB) being the leading infectious killer globally, our understanding of its pulmonary dynamics remains limited. We present an in silico approach to adapt a digital human lung framework in order to explore the dynamics of lesion encapsulation from internal membranes and of the infection spreading through the bronchial tree. First, we implement an algorithm that constructs a 3-dimensional Voronoi tessellation confined to a lung geometry to model secondary pulmonary lobules, enabling control of the distribution of sizes and positions of the cells. This results in anatomically accurate volumes and allows the generation of virtual populations of different ages and morphological characteristics. Next, we simulate infectious aerosol transport in the first generations of an anatomically accurate bronchial tree geometry using Navier-Stokes-based computational fluid-particle dynamics, accounting for particle-wall interactions and a dynamic breathing cycle. We consider an initial infection of TB and the airborne endogenous reinfection process to obtain deposition maps and travelled distance distributions that tie in with existing literature but suggest a necessary improvement of the boundary conditions. These probabilities continue in the direction of producing quantitative results to implement directly on the Bubble Model, a framework that describes lesion dynamics on a macroscopic level.