The Creation and Statistical Evaluation of a Deterministic Model of the Human Bronchial Tree from HRCT Images

Abstract

A quantitative description of the morphology of lung structure is essential prior to any form of predictive modeling of ventilation or aerosol deposition implemented within the lung. The human lung is a very complex organ, with airway structures that span two orders of magnitude and having a multitude of interfaces between air, tissue and blood. As such, current medical imaging protocols cannot provide medical practitioners and researchers with in-vivo knowledge of deeper lung structures. In this work a detailed algorithm for the generation of an individualized 3D deterministic model of the conducting part of the human tracheo-bronchial tree is described. Distinct initial conditions were obtained from the high-resolution computed tomography (HRCT) images of seven healthy volunteers. The algorithm developed is fractal in nature and is implemented as a self-similar space sub-division procedure. The expansion process utilizes physiologically realistic relationships and thresholds to produce an anatomically consistent human airway tree. The model was validated through extensive statistical analysis of the results and comparison of the most common morphological features with previously published morphometric studies and other equivalent models. The resulting trees were shown to be in good agreement with published human lung geometric characteristics and can be used to study, among other things, structure-function relationships in simulation studies.

Anthropometry‐based generation of personalized and population‐specific human airway models

“…The current approach is different from existing personalization models in the following ways: (a) It can use different anatomical measurable dimensions, unlike the standard height and the weight parameters, employed by many researchers and (b) person‐specific lung morphological template can be used (if known), instead of the standard template (Zygote, as in this paper). Hence, making it more person specific.…”

Predicted airway obstruction distribution based on dynamical lung ventilation data: A coupled modeling‐machine learning methodology

“…From 4 scans, upper airways can be segmented and used along with the parenchyma surface to propagate a conductive tree model with a patient-specific space filling approach [36], see Figure 7. The resulting structure contains a faithful representation for segmented upper branches, and a simulated tree representation model for the remaining airways.…”

“…The resulting left lung tree used herein contains 954 airways among which 477 are terminal branches. When simulating asthma, we generate {±BC²Pv plugs with {±BC²Pv randomly chosen in the interval ⟦1; 40⟧ (see subsection 2.1), this interval being set based on non-published data from [36].…”

Calculated ventilation and effort distribution as a measure of respiratory disease and Heliox effectiveness

“…The segmented upper airways and lung envelope were the initial conditions used to build a lobar space-filling tree following the methodology described in (Montesantos et al, 2016). From the lung envelope a 3D mesh is built using Meshlab (Meshlab -developped with the support of 3D-CoForm project -meshlab.sourceforge.net, n.d.) and Gmsh (Geuzaine and Remacle, 2009).…”