• Libraries
    • Log in to:
    View Item 
    •   MSpace Home
    • Faculty of Graduate Studies (Electronic Theses and Practica)
    • FGS - Electronic Theses and Practica
    • View Item
    •   MSpace Home
    • Faculty of Graduate Studies (Electronic Theses and Practica)
    • FGS - Electronic Theses and Practica
    • View Item
    JavaScript is disabled for your browser. Some features of this site may not work without it.

    Snoring sounds analysis: automatic detection, higher order statistics, and its application for sleep apnea diagnosis

    Thumbnail
    View/Open
    Azarbarzin_Ali.pdf (2.073Mb)
    Date
    2011
    2012
    Author
    Azarbarzin, Ali
    Metadata
    Show full item record
    Abstract
    Snoring is a highly prevalent disorder affecting 20-40% of adult population. Snoring is also a major indicative of obstructive sleep apnea (OSA). Despite the magnitude of effort, the acoustical properties of snoring in relation to physiological states are not yet known. This thesis explores statistical properties of snoring sounds and their association with OSA. First, an unsupervised technique was developed to automatically extract the snoring sound segments from the lengthy recordings of respiratory sounds. This technique was tested over 5665 snoring sound segments of 30 participants and the detection accuracy of 98.6% was obtained. Second, the relationship between anthropometric parameters of snorers with different degrees of obstruction and their snoring sounds’ statistical characteristics was investigated. Snoring sounds are non-Gaussian in nature; thus second order statistical methods such as power spectral analysis would be inadequate to extract information from snoring sounds. Therefore, higher order statistical features, in addition to the second order ones, were extracted. Third, the variability of snoring sound segments within and between 57 snorers with and without OSA was investigated. It was found that the sound characteristics of non-apneic (when there is no apneic event), hypopneic (when there is hypopnea), and post-apneic (after apnea) snoring events were significantly different. Then, this variability of snoring sounds was used as a signature to discriminate the non-OSA snorers from OSA snorers. The accuracy was found to be 96.4%. Finally, it was observed that some snorers formed distinct clusters of snoring sounds in a multidimensional feature space. Hence, using Polysomnography (PSG) information, the dependency of snoring sounds on body position, sleep stage, and blood oxygen level was investigated. It was found that all the three variables affected snoring sounds. However, body position was found to have the highest effect on the characteristics of snoring sounds. In conclusion, snoring sounds analysis offers valuable information on the upper airway physiological state and pathology. Thus, snoring sound analysis may further find its use in determining the exact state and location of obstruction.
    URI
    http://hdl.handle.net/1993/9593
    Collections
    • FGS - Electronic Theses and Practica [25061]

    DSpace software copyright © 2002-2016  DuraSpace
    Contact Us | Send Feedback
    Theme by 
    Atmire NV
     

     

    Browse

    All of MSpaceCommunities & CollectionsBy Issue DateAuthorsTitlesSubjectsThis CollectionBy Issue DateAuthorsTitlesSubjects

    My Account

    Login

    Statistics

    View Usage Statistics

    DSpace software copyright © 2002-2016  DuraSpace
    Contact Us | Send Feedback
    Theme by 
    Atmire NV