Simulation of respiratory flow in human upper airway model

Abstract

Obstructive sleep apnea (OSA) is a common disorder characterized by partial or complete narrowing of the pharyngeal airway during sleep. The pathogenesis of this disorder is not, however, fully understood yet, and a better understanding of OSA pathophysiology is required in order to guide treatment planning. Widening upper airway is a common surgery to treat severe OSA, but the success rate is quite low. To understand the pathogenesis of OSA from fluid mechanics point of view, we carried out both numerical and experimental investigations in this study. First of all, as Computational Fluid Dynamics (CFD) is a potential non-invasive tool for investigating the pathophysiology of OSA, we carried out CFD for numerical simulation for both normal and OSA subjects. To build the idealized model for CFD simulation, the medical imaging technique is used to generate the upper airway model. Secondly, because the flow in the upper airway region is expected to be turbulent, different turbulence models including Unsteady Average Navier-Stokes (URANS), two-equation turbulent models (unsteady k-ε, standard k-ω, and k-ω Shear Stress Transport) and Large Eddy Simulation (LES) model were compared in CFD numerical simulation. It is concluded that the LES model should be the most appropriate model for this CFD simulation of OSA. To validate the suitability of CFD modeling methods, we carried out Laser Doppler measurement in 3D-printing OSA upper airway models, and found excellent agreement between the measured and calculated velocity profiles in two upper airway models for the first time. Then four pairs of OSA upper airway subjects with 8 different apnea-hypopnea index(AHI) values are investigated with LES model. It is found that a dominant recirculation downstream of the minimum cross-section should be a main feature of a successful surgery, and the strength of 3-5 Hz signal induced by flow separation in the upper airway plays an important role in appraising breathing quality. This provides a new guideline for surgery planning. Finally, the stochastic resonance(SR) phenomenon was investigated in both normal and OSA subjects. We found that the SR phenomenon is existed in both subjects. The strong correlation between the signal-to-noise ratio (SNR) and AHI indicates that SR may play an important role in the respiratory system as periodic oscillating signals are enhanced significantly by noise. It seems that the quality of the oscillating signal can serve as a quantitative measure to quantify the breathing quality of OSA subject.