Relationship between Swallowing Ability and Frequency Features of Swallowing Sounds Collected by Using General-Purpose Instrument

Abstract

  This study aimed to clarify the relationship between swallowing ability assessment by screening tests and the frequency features of swallowing sounds collected using a commercially available, inexpensive, general-purpose neck-mounted microphone and smartphone. Thirty-three hospitalized patients underwent modified water swallowing test (MWST) and repetitive saliva swallowing test (RSST) and were divided into two groups according to their MWST (declined: ≤ 3, normal: ≥ 4) and RSST (declined: ≤ 2, normal: ≥ 3) scores. The pharyngeal sounds were recorded on a smartphone equipped with a neck-mounted microphone. A two second period of pharyngeal sounds centered on the volume peak was extracted as the swallowing sound, multiplied by a Hanning window, and analyzed using a discrete Fourier transform to extract the amplitude spectra. Four frequency feature quantities, namely spectral centroid, spectral variance, spectral flatness, and peak frequency, were calculated from each amplitude spectrum. The results of the Mann-Whitney U test performed between the two groups indicated significant differences in spectral variance, spectral flatness, and peak frequency in the high-frequency bands in the RSST and in spectral variance in the all-frequency bands in the RSST. A significant difference in the frequency features of the swallowing sounds depending on the swallowing ability in the general-purpose instrument configuration was indicated.