PLoS ONE (Jan 2021)

Human upper-airway respiratory airflow: In vivo comparison of computational fluid dynamics simulations and hyperpolarized 129Xe phase contrast MRI velocimetry.

  • Qiwei Xiao,
  • Neil J Stewart,
  • Matthew M Willmering,
  • Chamindu C Gunatilaka,
  • Robert P Thomen,
  • Andreas Schuh,
  • Guruprasad Krishnamoorthy,
  • Hui Wang,
  • Raouf S Amin,
  • Charles L Dumoulin,
  • Jason C Woods,
  • Alister J Bates

DOI
https://doi.org/10.1371/journal.pone.0256460
Journal volume & issue
Vol. 16, no. 8
p. e0256460

Abstract

Read online

Computational fluid dynamics (CFD) simulations of respiratory airflow have the potential to change the clinical assessment of regional airway function in health and disease, in pulmonary medicine and otolaryngology. For example, in diseases where multiple sites of airway obstruction occur, such as obstructive sleep apnea (OSA), CFD simulations can identify which sites of obstruction contribute most to airway resistance and may therefore be candidate sites for airway surgery. The main barrier to clinical uptake of respiratory CFD to date has been the difficulty in validating CFD results against a clinical gold standard. Invasive instrumentation of the upper airway to measure respiratory airflow velocity or pressure can disrupt the airflow and alter the subject's natural breathing patterns. Therefore, in this study, we instead propose phase contrast (PC) velocimetry magnetic resonance imaging (MRI) of inhaled hyperpolarized 129Xe gas as a non-invasive reference to which airflow velocities calculated via CFD can be compared. To that end, we performed subject-specific CFD simulations in airway models derived from 1H MRI, and using respiratory flowrate measurements acquired synchronously with MRI. Airflow velocity vectors calculated by CFD simulations were then qualitatively and quantitatively compared to velocity maps derived from PC velocimetry MRI of inhaled hyperpolarized 129Xe gas. The results show both techniques produce similar spatial distributions of high velocity regions in the anterior-posterior and foot-head directions, indicating good qualitative agreement. Statistically significant correlations and low Bland-Altman bias between the local velocity values produced by the two techniques indicates quantitative agreement. This preliminary in vivo comparison of respiratory airway CFD and PC MRI of hyperpolarized 129Xe gas demonstrates the feasibility of PC MRI as a technique to validate respiratory CFD and forms the basis for further comprehensive validation studies. This study is therefore a first step in the pathway towards clinical adoption of respiratory CFD.