Abstract
Objective: OSA is caused by soft tissue collapse of the upper airway; treatment strategies can be determined depending on accurately localizing the obstruction sites. Acoustic analysis of snoring might provide a valuable way to localize the obstruction, if the association between the acoustics and the dynamics of the upper airway during OSA can be identified.
Method: Nine PSG OSA patients were enrolled. Dynamics of upper airway and the acoustic manifestations of snore events during natural sleep were recorded with dynamic MRI and a fiber-optic microphone simultaneously. The Gabor/wavelet transformation was applied to extract the acoustic representation of OSA events by removing the repeated-appearing MRI scanning background noises.
Results: All of the 9 patients were men (mean age, 48.3 ± 9.9 years, mean BMI, 25.1 ± 1.9). The snore events were categorized into typical (velopharyngeal) and mixed (combined with hypopharyngeal) obstructions. The mean frequency (161 ± 17.4 vs 188 ± 24.1 Hz; P < 10-16) and high (151~350 Hz)/low(80~150 Hz) frequency power ratio (1.08 ± 0.31 vs 1.55 ± 0.42 Hz; P < 10-16) of the snore sounds were significantly higher in mixed obstructions. However, the dynamic patterns of the frequency modulation varied with structural changes identified by dMRI. During the snore events, different characteristics of shifts in frequency were found (mean frequency changes: −28.3 ± 21.4 in typical obstruction vs 12.5 ± 22.9 Hz for mixed obstruction; P < 10-32), which gives better discrimination power.
Conclusion: The dynamics of acoustic patterns derived by Gabor spectrogram are unique and characteristics in different types of obstruction with the advantage of low cost and easy availability. Therefore, the dynamics of acoustic patterns warrants further investigation in combination with instantaneous structural changes of the obstruction sites in OSA patients through dMRI.
Get full access to this article
View all access options for this article.