COMPUTATIONAL AND EXPERIMENTAL MODELLING OF THE HUMAN UPPER AIRWAY ACOUSTICS WITH APPLICATION TO OBSTRUCTIVE SLEEP APNEA DIAGNOSIS DURING WAKEFULNESS

Obstructive Sleep Apnea (OSA) is a highly prevalent sleep-related breathing disorder characterized by repetitive upper airway collapse during sleep. Despite its clinical significance, current diagnostic techniques remain resource-intensive and often inaccessible. This thesis proposes two modelling approaches aimed at characterizing upper airway acoustics and investigates their potential application as non-invasive, quick and reliable OSA diagnosis during wakefulness. The first approach involves a hybrid aero-acoustic simulation framework that decouples airflow dynamics and sound propagation. Using patient-specific airway geometries reconstructed from high-resolution CT scans, Large Eddy Simulation (LES) was employed to model unsteady inspiratory flow. Aero-acoustic sources were extracted using Lighthill’s analogy and mapped onto a finite element mesh to solve the wave equation for acoustic propagation. This enabled detailed analysis of how anatomical features—including velopharyngeal, oropharyngeal, and tracheal constrictions—alter airflow patterns, acoustic pressure distribution, and resonance frequencies at the trachea. The results demonstrated strong correlations between geometrical narrowing and frequency shifts in tracheal breathing sound spectra. The second approach develops an empirical acoustic transfer function estimation between the mouth and the trachea during controlled wakefulness breathing. Participants were exposed to broadband and chirp input signals at the mouth, and tracheal responses were recorded to estimate the transfer function. A novel piezoelectric contact microphone was developed to record the sounds with minimal background interference. Statistical analysis revealed repeatable and individualized transfer functions, with significant spectral variations corresponding to anatomical differences and potential airway obstruction. Together, these approaches provide a comprehensive understanding of the relationship between the upper airway anatomy, airflow mechanics, and the spectral content of breathing sounds. The findings support the feasibility of using tracheal acoustic responses during wakefulness for OSA diagnosis. In addition, they provide mechanistic insights into OSA pathophysiology by linking specific acoustic signatures to underlying anatomical abnormalities. This enables not only OSA classification, but also identification of the likely anatomical contributors to obstruction in each case. Such individualized insights advance the understanding of the disorder mechanism and pave the way for more targeted, personalized therapeutic strategies. Furthermore, the methodology establishes the groundwork for developing accessible, non-invasive, and personalized screening tools for sleep-disordered breathing.