Evaluation of eustachian tube function using a portable device

The eustachian tube (ET) is a tubular organ that connects the tympanic cavity to the nasopharynx. The ET equalizes the air pressure in the tympanic ventricle with the atmospheric pressure and discharges secretions produced in the tympanic cavity into the pharynx. The ET is normally closed and opens during swallowing, but some people have symptoms, such as deafness, tinnitus, or autophony, caused by ET dysfunction (ETD), which was categorized by Smith et al.¹ as obstructive ET dysfunction or patulous eustachian tube (PET) dysfunction. Eustachian tube dysfunction has been reported to occur in approximately 1% to 5% in adults,^2–4 but is more common among children, and 40% of children under 10 years of age experience temporary ETD.⁴ 

The tubo-tympanoaerodynamicgraphy method, the tympanic impedance method, sonotubometry, and inflation-deflation tests have been proposed as methods by which to understand the physiological function of the ET. All-in-one measurement devices capable of performing various measurement methods are usually used in medical treatment. However, such conventional testing equipment is large and not portable, making it difficult to perform easy measurement and diagnosis. Moreover, these devices are expensive and so are difficult for patients to purchase. As such, hospitals are generally visited in order to perform these tests. On the other hand, since symptoms caused by ETD occur at various times in daily life, it is often difficult for patients to immediately visit a hospital for an examination when such symptoms suddenly occur. Smith et al. stated¹ that there is a need for new testing equipment that can indicate whether a patient has a tendency to develop ETD intermittently, even if the patient is free of symptoms at the time of testing. Although it is important to be able to quickly and easily test ET function (ETF) using portable media in cases such as those mentioned above, such a system has not yet been proposed.

Among the above-mentioned methods, sonotubometry is a relatively simple method, in which acoustic signals are emitted to the nasal cavity by a loudspeaker and the ET opening can be detected using the signals transmitted to the ear canal recorded by microphones. This method can be used with or without perforation of the eardrum, and the load on the human body is small, so, using this method, it is possible to measure the ETF of not only adults but also children and elderly persons. Sonotubometry, as mentioned above, is an easy-to-use method, but requires the use of a large loudspeaker with high output power to measure the sound transmitted to the ear canal with a sufficient signal-to-noise ratio, which is a crucial factor that reduces its versatility. Sonotubometry was developed by Perlman,⁵ and its accuracy, including reproducibility, has been verified in recent years.⁶ Compared to the above-mentioned conventional ETF measurement adopting broadband sound signals, some research investigates ETF measurement based on the pure-tone signals⁷ that has a possibility of providing enough output power with even much smaller speakers than those used for the conventional measurement. However, the reliability of the measurement by a pure tone including a frequency component at a single frequency of 8 kHz is concluded to be less reliable than that by a broadband sound due to noise pollution.⁷ Another measurement method using a novel click stimulus that is a sequence of square pulses and having broadband frequency component has been proposed.⁸ This study focuses on a more efficient broadband signal that can distinguish between normal and patulous ET subjects. They use smaller speakers compared to those utilized in the conventional test. However, in this study, the portability of the system itself is not discussed. Although stationary test equipment based on the above-mentioned methods has already been considered, test equipment with portability that can be used for testing at home or wherever they have not yet considered.

In the present study, we developed a portable ETF testing device using the sonotubometry of pure-tone driving by a portable device. This system enables the sound reproduction, recording, and signal processing to obtain the transmitted sound via ET, all with just a portable Android device and a lightweight speaker and microphone attached. By using pure-tone drive, the system is able to produce as high power as possible with a small speaker, but in order to eliminate subject selectivity as much as possible, the output signal made by superposing frequency components at two frequencies. In order to confirm the acoustic performance of the device, the ETFs for adults were measured by the proposed device, and the obtained results were compared with the results measured by the existing testing device.

Sonotubometry is used as a diagnostic method to evaluate the state of the ET based on the extent of sound transmission via the ET. First, a test sound is output from a loudspeaker into the nasal cavity. Then, the sound transmitted through the ET into the ear canal can finally be recorded by a microphone placed inside the ear canal. The ET is generally closed, but the ET opens when the subject swallows. In the case of a healthy subject, the ET immediately closes a short time after opening of the ET. The opening and closing of the ET can be detected by the increase and decrease in the sound pressure level of the transmitted sound. The ETF of the subject can be diagnosed based on the shape of the level waveform. In the present state of the art, some types of sound are used as test sounds. In most cases, a pure tone or band noises in the frequency range of approximately 7 to 8 kHz is used.^7,9 However, in other cases,⁷ the measurement accuracy based on the pure-tone test has been reported to be insufficient to diagnose ETD. They discuss the statistical characteristics of the opening time of ET and the increased sound level in the timing of ET opening and concluded that the 8 kHz pure tone is not reliable since some of the measurement failed to record an opening of ET despite the fact that all of the subjects reported that an opening. However, in this study, they have not yet compared their measured results with those by other methods using such broadband sound sources, which is also necessary to investigate the reliability of the proposed method. The other cases use band noises having a center frequency of approximately 7 kHz.^10–12 The rate of opening detection of ET has been widely distributed from 60% to 90% in many cases.^10,13,14 The main reason for deterioration of the detection accuracy has been attributed to the decrease in the sound-to-noise ratio due to physiological noise.¹¹ 

In order to efficiently and easily obtain the daily fluctuation of the ETF, a portable device was adopted as the control device of the ETF measurement. However, the output level of the sound from a smaller speaker attached to a portable device is not high compared to that of a relatively larger speaker used in conventional commercial equipment. Under such a situation of the hardware settings, the sound signal appropriate for the portable measurement should be chosen from the acoustic feature of the broadband and narrowband sounds as follows. First, the broadband sound enables one to stably measure the transmitted sound via ET even if the frequency at which the test sound can easily transmit via ET slightly fluctuates due to individual difference of ET shape, while the reproduction at a certain sound pressure level that enables one to obtain sufficient signal-to-noise ratio requires relatively large loudspeakers to output relatively high sound energy of the broadband noise. In contrast to this, the reproduction of narrowband sound at a certain sound pressure level that enables one to obtain sufficient signal-to-noise ratio requires relatively smaller speakers because the narrowband sound has relatively low sound energy compared to that of the broadband sound, while the transmission characteristics of some subjects via ET may be degraded when the frequency at which the test sound can easily transmit slightly deviates. By considering the above pros and cons of the broadband and narrowband sounds, it can be said, in this case, to be more appropriate to use pure-tone driving than broadband driving to provide the user comfortable portability, as long as the appropriate frequency components of the pure-tone test signal that can efficiently detect the sound transmission via ET is found. Therefore, in the present study, a pure tone was adopted as the test sound. The details of the portable device and the frequencies at which the test sound was generated are described below.

First, the proposed apparatus and detail flow of sound reproduction and recording are shown in Figs. 1(a) and 1(b), respectively. As shown in Fig. 1(a), this system consists of a portable device for processing the audio input and output signals, a speaker and a microphone for reproduction and recording of the audio signals, and a USB audio interface for connecting the portable device and the speaker and microphone. This device completely works as a stand-alone device. Since this system is quite lightweight and does not require a power supply, the user can measure the ETF everywhere with this fully portable system. Next, as shown in Fig. 1(b), the test sound is reproduced by the speaker. The sound transmitted via the ET, the middle ear cavity, and the ear canal is then recorded by the microphone. The Android device is required to simultaneously reproduce the test sound and record the transmitted sound. However, analog input and output from the earphone jack cause a ground loop, which prevents accurate measurement. Therefore, a USB audio interface based on the USB On-The-Go standard, which is an extension of the USB 2.0 standard, is adopted to connect the speaker, the microphone, and the Android device. As the USB audio interface, Sound Blaster PLAY! 3 (Creative), which does not require an external power supply, is used, and A/D conversion is performed for digital input and output to the Android device.

In order to examine the ETF by sonotubometry, it is necessary to insert a speaker into the nasal cavity and a microphone into the ear canal. Therefore, the speaker and the microphone should be small enough to fit into the ear and nose. The microphone used was an ME15 [Olympus, Figs. 1(a) and 1(c)]. The microphone has omni-directional characteristics with a frequency response from 100 to 12 kHz and a sensitivity of −42 dB at 1 kHz (0 dB = 1 V/Pa). The external dimensions of the microphone are a diameter of 8.8 mm and a length of 34.3 mm, and the mass of the microphone is 11 g. The speaker used was an ATH-CK330M [Audio Technica, Figs. 1(a) and 1(d)]. The diameter of the speaker driver is 10 mm. The frequency range of the speaker is 20 to 23 000 Hz, and the impedance of the speaker is 24 Ω. In order to control these devices, an Android device (SO-03F, SONY), which is popular and relatively inexpensive, was used. Related to the sustainability of this system, it is convenient for the user that the microphone and speaker inserted into the ear and nose have disposable covers. The tip part of the speaker that is inserted into the nose is detachable, and this part can be replaced with a new attachment that is commercially available. In contrast, the microphone is used in a genuine condition without any detachable covers in the experiment of this study. However, it is possible to make a detachable cover made of thin and flexible rubber material that does not affect the acoustical characteristics of the microphone itself. Since the diameter of the microphone is slightly smaller compared to that of the opening part of the ear canal, it is possible to insert the microphone covered by such a rubber cover. Therefore, the investigated system in this study can be sustainable.

Next, the method by which a sonotubogram is obtained, as shown in Fig. 2(c), is described. First, the transmitted sound of the reproduced sound into the ear canal was recorded as shown in Fig. 2(a). In this measurement, the subject swallowed twice in the timing indicated by arrows, and ET is opened twice in this timing. But any transmitted sound was clearly shown in Fig. 2(a). This is because the recorded sound is polluted by the physiological noise. To remove the noise, the recorded sound should be filtered by a bandpass filter that filters the frequency components at the frequencies of the test sound. The example of the filtered sound wave p(t) is shown in Fig. 2(b). The transmitted sounds are indicated by two arrows in the timing of swallowing twice. Then, p(t) is converted to a sonotubogram [Fig. 2(c)] as 10log₁₀p²(t). In Fig. 2(c) two peaks that indicate the sound transmission via ET can be seen. Based on each of the shapes of the peaks, we can discuss ETF. Note that the frequency components of the test sound will be determined in Sec. 2.3. A PC with a stationary audio interface (Roland, Octa Capture), as shown in Fig. 1(e), was used instead of the Android device with a USB audio interface only in Sec. 2.3 to determine the frequency components of the test sound.

In the sonotubometry test, a test sound is reproduced from a speaker inserted into the nasal cavity, and the sound propagates through an opening in the pharynx into the ET. In addition, the transmitted sound is finally received in the ear canal through the middle ear. The sound incident from the pharynx into the ET is greatly attenuated when the sound passes through the ET, so it is necessary to record the transmitted sound with a sufficient signal-to-noise ratio with respect to the background noise inside the ear canal. In conventional sonotubometry, a stationary and large speaker with a high output level is inserted into the nasal cavity, which radiates the sound source signal into the nasal cavity. In addition, due to the higher output level of the large speaker, the conventional scheme could keep the sound-to-noise ratio higher even if a band noise was adopted in the sound reproduction. In contrast, the adopted compact speaker does not have a very large output sound power. In order to record transmitted sound with the highest possible signal-to-noise ratio when using the compact speaker, it is important to reproduce the total sound energy concentrated at the narrowest band possible, while the adoption of a narrowband tone, such as a pure tone, may reduce the frequency selectivity of sound transmission via the ET. For the above reason, a test sound with tonal frequency components was adopted.

In order to determine the tonal components of the test sound, the sound transmission characteristics through the ET was measured using the time-stretched-pulse (TSP) signal when the ET was open and closed, respectively. Herein, 2¹⁷ samples of sound data at 48 kHz sampling were used as the TSP signal sweeping from 0 Hz to 24 kHz in 2.73 s. The sound transmission characteristics via the closed ET were measured in the natural state of the subject, while those via the open ET were measured when the ET of the subject is continuously opening using Valsalva maneuver during the playback time (2.73 s) of the TSP signal.

As a basic study, the experimental results for one subject are shown in Fig. 3. Figure 3(a) shows the frequency characteristics of the sound transmitted to the microphone. The result indicates that a significant sound transmission via the ET was observed in the band from 6 to 12 kHz indicated by gray in the figure, while the sound pressure level (SPL) under 6 kHz indicates relatively similar values between the case of an open ET and a closed ET due to the physiologically radiated noise.¹¹ Next, in order to investigate the sound transmission characteristics in the band from 6 to 12 kHz in more detail, the measurement was performed again using a band limited noise sound with only the 6- to 12-kHz-band component. The results of Fig. 3(b) show that a relatively higher-level difference between the transmitted sound levels of the open ET and the closed ET at each of the frequencies from (1) to (5). Then, pure-tone signals at frequencies (1) to (5) (6.4, 7.0, 7.7, 9.5, and 11.8 kHz) were generated, and the transmitted SPL were again measured using each of these signals. As a result, there was a difference in the transmitted sound level when the ET was closed or opened at frequencies (2) to (5). However, the main purpose of the present study was not to obtain a detailed understanding of the transmitted sound characteristics through ET, but rather to realize a portable device for efficient ETF testing. Thus, the frequency component of the test sound was determined to be 9.5 kHz, at which the level difference was the largest, and 7 kHz, at which the level difference is not large, but the conventional sonotubometry adopts this frequency component. Note that the frequency characteristics of sound transmission have some possibility to fluctuate depending on each of the subjects. However, small fluctuation in transmission frequency will not have much of an effect on the measured results because the peak at 9.5 kHz has relatively broad width compared to other peaks and the SPL difference between the situations of ET open and close shown in Fig. 3(b) indicates a relatively large value especially over 8 kHz. Besides, the usage of two frequency components will largely reduce the uncertainty of the measured result. The validity of the method using these two pure tones is verified in Sec. 3.

The determined test signal including the frequency components at 7 and 9.5 kHz was reproduced via the system based on Android by SONY shown in Figs. 1(a) and 1(b). The signal was made as the maximum amplitude of the periodic signal with 7 and 9.5 kHz components (16 bit sampling) is 32 768, and was reproduced by the Android device with its maximum volume of the audio output. The output sound pressure level by the JK-05A (RION) that will be compared with the present study in Sec. 3 was also measured. The frequency characteristics of each of the output sound pressure levels are indicated in Fig. 3(d). Note that both of these measurements were performed by inserting the speaker into the opening of one side of a rubber tube with a length of 10 cm and a diameter of 0.7 cm, and the transmitted sound inside the tube was recorded by the microphone (ME15, Olympus) inserted into the other opening. This tube simulates the space inside the nasal cavity. The band noise radiated by the system of RION has quite large energy compared to the pure tone by the present system since the former system uses a quite large loudspeaker whereas the latter one uses a small speaker.

The ETF was measured in 12 ears of six subjects with no subjective symptoms of the ETF using the JK-05A, which is an existing ETF test device that is frequently used in actual medical practice. In addition, the ETF was also measured in the same 12 ears using the proposed portable device. The measurement was performed twice for each of the subjects. Thus, a total of eight sonotubograms for each of the subjects were obtained.

In the measurement using the portable device, the sampling frequency was 48 kHz, and the test sound was a superimposed signal of two pure tones at 7 kHz and 9.5 kHz. As already mentioned in Sec. 2, in order to remove the physiological noise, the frequency components only at 7 kHz and 9.5 kHz were filtered out. Note that, in the present study, opening of the ET was provoked by dry swallowing,⁹ where both dry swallowing and water swallowing contribute to the reliability of the measurement to the same extent.

In this validation study, two quantities, the ET opening time (T_open) and the sound pressure difference in the ear canal between the cases of an open ET and a closed ET (ΔL), are defined as shown in Fig. 4(a), and the results measured by the existing and proposed methods are compared. In order to obtain these quantities, the maximum sound pressure level L_max of the peak shape generated during swallowing was determined. At the same time, the average sound pressure level of the background noise, L_BGN,ave, was determined, and the difference between these levels, ΔL, was calculated. After that, intersection points P1 and P2 at t₁ and t₂ between the vertical line at L_BGN,ave and the measured line were determined, where T_open = t₂ − t₁.

An example of the measurement results for one subject is shown in Fig. 4(b). Here, the characteristics of T_open for the right ear indicate a relatively longer duration, as compared to those for the left ear. Although this tendency can be observed in the results obtained by both devices, the characteristics of ΔL obtained by the portable device are lower than those obtained by the existing device. This tendency may be affected by the lower signal-to-noise ratio due to the smaller output of the sound energy from the portable device.

The values of T_open and ΔL measured by each of the portable device and the existing device (JK-05A) are comparatively plotted in Figs. 4(c) and 4(d), respectively. Note that ET opening was confirmed in all trials of dry swallowing, and all of the ET openings (24 times in total for all of the subjects) were detected as a peak waveform of sonotubometry, as shown in Fig. 4(b). In these figures, the regression equations and correlation coefficients are shown. The results for both T_open and ΔL have correlation coefficients of over 0.8, whereas the gradients of the regression equations have values smaller than 1.0. In particular, the results for ΔL have relatively smaller gradients. These results are affected by the relatively lower signal-to-noise ratio in the case of the portable device, as compared to that of the existing device. The lower gradient in the case of T_open is also due to the lower signal-to-noise ratio of the portable device. However, it has been shown that the proposed device can be used to obtain ΔL in a range up to approximately 20 dB and T_open up to approximately 1.2 s. In addition, in the present study, T_open and ΔL were, respectively, obtained as 528 (SD: ±256) ms and 11.8 (SD: ±3.4) dB by the portable device and 660 (SD: ±326) ms and 17.1 (SD: ±8.9) dB by the existing device. In previous research, it was reported that the mean ET opening duration times corresponding to T_open were measured using healthy subjects as 270 (SD: ±96) ms,⁹ 670 (SD: ±324) ms,¹⁵ and 363 (SD: ±192) ms,¹⁶ and the mean sound wave amplitude corresponding to T_open were measured using healthy subjects as 13.48 (SD: ±6.57) dB,⁹ 17.2 (SD: ±4.09) dB,¹⁵ and 16.86 (SD: ±6.16) dB,¹⁶ respectively. Note that, among these cases,^9,15,16 the broadband test sound is used. The values of T_open and ΔL measured by the proposed portable device generally correspond with the values in previous research. Although the low output level from the speaker of the proposed portable device may cause a slight decrease in the results for ΔL, it has been suggested that the present results using the portable device with pure-tone test sounds can evaluate the T_open and ΔL with sufficient accuracy to grasp the tendency of the ETF.

The acoustic performance of a portable ETF testing device by sonotubometry based on pure-tone inspection was examined through comparison with the results of the existing testing technique based on broadband sound inspection. The measurement results for T_open and ΔL using pure-tone sounds including the frequency components at two frequencies of 7.0 and 9.5 kHz showed generally appropriate correspondence from the viewpoint of daily and casual inspection of the ET for patients with intermittent symptoms.