Elemental Research for Quantification of Eustachian Tube Insufflation Method

Commercial equipment is available for conducting Eustachian tube function tests but, generally, the Eustachian tube insufflation method doubles as a treatment and diagnosis. However, with the Eustachian tube insufflation method, the doctor merely performs the auscultation. The Eustachian tube insufflation sounds are not saved as data, and the diagnosis relies on the auditory judgment of the doctor. That is, the technique is not objective. To overcome these disadvantages, we set out to develop a system that would enable the objective evaluation of the Eustachian tube insufflation sounds. We sampled a wide range of Eustachian tube insufflation sounds and tried to identify the acoustic features of specific pathological conditions. However, it was so difficult to find the features of Eustachian tube insufflation sounds and to classify according to observation of spectra completely. Therefore, we proposed a method of inputting the frequency analysis results as feature vectors to a Self-Organizing Map (SOM). Our results confirmed that, when the maximum values of the spectra are normalized using the 0to 5-kHz frequency analysis results, the normal sounds, stenosis, and crepitation could be classified.