Automated Acoustic Analysis in Detection of Spontaneous Swallows in Parkinson’s Disease
- 515 Downloads
Acoustic monitoring of swallow frequency has become important as the frequency of spontaneous swallowing can be an index for dysphagia and related complications. In addition, it can be employed as an objective quantification of ingestive behavior. Commonly, swallowing complications are manually detected using videofluoroscopy recordings, which require expensive equipment and exposure to radiation. In this study, a noninvasive automated technique is proposed that uses breath and swallowing recordings obtained via a microphone located over the laryngopharynx. Nonlinear diffusion filters were used in which a scale-space decomposition of recorded sound at different levels extract swallows from breath sounds and artifacts. This technique was compared to manual detection of swallows using acoustic signals on a sample of 34 subjects with Parkinson’s disease. A speech language pathologist identified five subjects who showed aspiration during the videofluoroscopic swallowing study. The proposed automated method identified swallows with a sensitivity of 86.67 %, a specificity of 77.50 %, and an accuracy of 82.35 %. These results indicate the validity of automated acoustic recognition of swallowing as a fast and efficient approach to objectively estimate spontaneous swallow frequency.
KeywordsSwallowing assessment Acoustic Spontaneous swallow Swallow frequency Deglutition Deglutition disorders
Conflict of interest
The authors have no conflicts of interest to disclose.
- 6.Yadollahi A, Moussavi ZM. A model for normal swallowing sounds generation based on wavelet analysis. 2008 CCECE, 21st IEEE Canadian Conference on Electrical and Computer Engineering, Niagara Falls, Ontario, Canada, 4–7 May 2008.Google Scholar
- 9.Makeyev O, Sazonov E, Schuckers S, Lopez-Meyer P, Baidyk T, Melanson E, Neuman M. Recognition of swallowing sounds using time-frequency decomposition and limited receptive area neural classifier. Appl Innov Intell Syst. 2008;XVI:33–46.Google Scholar
- 13.Aboofazeli M, Moussavi Z. Automated extraction of swallowing sounds using a wavelet-based filter. EMBS’06 28th Annual International Conference of the IEEE Engineering in Medicine and Biology Society, New York, August 31, September 3, 2006, vol 1, pp 5607–10.Google Scholar
- 14.Aboofazeli M, Moussavi Z. Automated classification of swallowing and breath sounds. EMBS’04 26th Annual International Conference of the IEEE Engineering in Medicine and Biology Society, Conference Proceedings: Linkages for Innovation in Biomedicine: 1–5 September 2004, San Francisco, vol 6, pp 3816–19.Google Scholar
- 15.Amft O, Troster G. Methods for detection and classification of normal swallowing from muscle activation and sound. Pervasive Health Conference and Workshops, Innsbruck, Austria, November 29, December 1, 2006, pp 1–10.Google Scholar
- 19.Mitchell TM. Machine learning. WCB. Boston: McGraw-Hill Science/Engineering/Math; 1997.Google Scholar
- 20.Weickert J. A review of nonlinear diffusion filtering. In: Scale-Space Theory in Computer Vision, First International Conference, Scale-Space ’97, Utrecht, The Netherlands, July 2–4, 1997, Proceedings (Lecture Notes in Computer Science). Berlin: Springer-Verlag, 1997, No. 1252, pp 1–128.Google Scholar
- 24.Martin-Harris B. Coordination of respiration and swallowing. GI Motility Online, 16 May 2006. doi: 10.1038/gimo10. Available at http://www.nature.com/gimo/contents/pt1/full/gimo10.html.
- 25.Golabbakhsh M.Tracheal breath sound relationship with respiratory flow: modeling the effect of age and airflow estimation. Dissertation, University of Manitoba, 2004.Google Scholar
- 27.Crary M, Carnaby-Mann G, Sura L, Sia I. Spontaneous swallow frequency as a screening protocol for dysphagia in acute stroke. Neurology. 2013;80(S32):004.Google Scholar