Introduction: The acoustic reflex is the active response of the middle ear to loud sounds, altering the mechanical transfer function of the acoustic energy into the inner ear. Our goal was to observe the effect of the acoustic reflex on the tympanic membrane by identifying a significant nonlinear increase in membrane oscillations. Methods: By using interferometric spectrally encoded endoscopy, we record the membrane oscillations over time in response to a loud, 200-ms-long acoustic stimulus. Results: A gradual reflex activation is measured between approximately 40 and 100 ms, manifested as a linear 42% increase in the umbo oscillation amplitude. Conclusion: The measured oscillations correlate well with those expected from a mechanical model of a damped harmonic oscillator, and the results of this work demonstrate the potential of interferometric spectrally encoded endoscopy to observe unique dynamical processes in the tympanic membrane and in the middle ear.

The instantaneous response of the tympanic membrane and the middle ear ossicles to a wide range of acoustic stimuli is known to be approximately linear, i.e., their oscillation amplitudes are linearly proportional to the applied sound pressures [1, 2]. When the sound levels cross a certain threshold, normally between 70 and 100 dB HL, the stapedial muscle contracts in an effect known as the acoustic reflex, or the middle ear muscle reflex [3‒7], regulating the sound energy that is transferred into the inner ear. As the acoustic reflex arc is mediated by the brainstem, this reflex can be used to diagnose cochlear and retro-cochlear disorders [8, 9]. Additionally, as the acoustic reflex is activated by the facial VII cranial nerve, disorders associated with this nerve may also be diagnosed.

Most commonly, the acoustic reflex is measured using a standard immittance test [10‒12], in which sound admittance into the middle ear is estimated by measuring the acoustic impedance with a 226 Hz probe tone, where a strong stimulus is applied to the same ear for triggering an ipsilateral reflex, or to the opposite ear for triggering a contralateral reflex. By varying the frequency of the probe tone, the measured sound admittance into the middle ear was found to be frequency-dependent [13], where the reflex reduces the admittance of probe frequencies below 800 Hz and increases the admittance at higher frequencies between 800 Hz and 2 kHz. Above 2 kHz, the effect of the reflex was found negligible [14, 15], whereas an admittance peak at approximately 1 kHz was reported by Feeney et al. [16].

The sound admittance into the middle ear could also be measured by means of optical interferometry, which allows sufficient sensitivity to directly measure nanometric vibrations. By interfering narrowband light reflections from the tympanic membrane with a reference signal, laser Doppler vibrometry (LDV) [17] was able measure the nanometric vibrations of a single point on the tympanic membrane, both in vitro [18, 19] and in vivo [20]. Recently, Jones et al. [21] have used LDV for studying whether the acoustic reflex can be triggered for preconditioning the ear to loud sounds. Other methods that could potentially be used for detecting the acoustic reflex may include phase-sensitive OCT [19, 22, 23] and stroboscopic holography [24, 25], as both technologies have been previously demonstrated capable of high-sensitivity vibration measurement.

By utilizing broadband light for encoding a single lateral dimension with wavelength, interferometric spectrally encoded endoscopy (ISEE) allows measuring nanometric tissue motion across an entire transverse line through a single optical fiber [26], whereas full 2D imaging of the vibrating surface could be accomplished by slowly scanning the spectrally encoded line across the tissue [27, 28]. Using a dedicated handheld imaging probe that could be easily inserted into the ear canal, ISEE has been demonstrated useful for imaging the pure tone and broadband response of the tympanic membrane in vivo [29, 30], providing detailed mechanical analysis of its various anatomical structures.

In this paper, we demonstrate in vivo imaging of the tympanic membrane during the activation of the acoustic reflex, visualizing its effect on the membrane. By conducting measurements at different sound levels, we show that while the full oscillation pattern of the membrane remains generally unchanged, the oscillation amplitudes across the membrane vary significantly from what is expected from a simple linear response. By focusing the measurement on the umbo (UM) region and applying an above-threshold acoustic square-envelope wave with 1-kHz carrier frequency, we find a gradual 42% increase in the vibration amplitude, which takes place only between 40 ms and 100 ms after the beginning of the sound stimulus. When fitting our results to the mechanical model of a damped harmonic oscillator, we find good correlation at the initial step response, throughout the gradual activation of the acoustic reflex, during the following steady-state oscillations, and at the final oscillation decay.

The experimental setup for in vivo imaging of the tympanic membrane vibrations during activation of the acoustic reflex is shown in Figure 1. Briefly, a transverse spectrally encoded line was generated by focusing broadband light from a polarization-maintaining SLD (50-nm bandwidth, 840-nm central wavelength) on the tympanic membrane, using a diffraction grating (1,200 lines/mm) and an achromatic lens system (5 lenses total). The sound stimulus was applied to the ear canal of a human volunteer (healthy male, 36 years-old, Helsinki approval protocol RMB-0634-17) through a dedicated port in the modified conventional otoscope at the distal end of the handheld system. Simultaneously with the acoustic stimulus, the transverse spectrally encoded line was scanned across the tympanic membrane using a galvanometric scanner, and the light reflections from the membrane were allowed to interfere with a reference beam on a high-speed spectrometer. The resulting spectral interferograms were recorded using a high-speed line camera and analyzed for extracting axial surface motion [27‒30]. The resulting raw data provide the full axial displacement period across the entire 2D field of view, with approximately 50-μm lateral resolution and up to 6-nm axial resolution, at an imaging rate of 2 frames per second. In the single-line measurement mode, the spectrally encoded line was placed at a selected line of interest and the acquisition speed was as high as 70 kHz, limited only by the line rate of the spectrometer camera. An additional widefield imaging channel was used to guide the measurement, comprised the integral white-light illumination of the otoscope and a high-speed (60 fps) color video camera.

Fig. 1.

Handheld ISEE system for measuring the acoustic reflex. Near-infrared light from a polarization-maintaining SLD (PM-SLD) was directed through a 50:50 fiber coupler to a reference arm and to the ISEE-otoscope system designed to illuminate the tympanic membrane with the transverse spectrally encoded line. A high-speed spectrometer was used for measuring the interference between the sample and reference arms, and the results were analyzed by a personal computer, which also controlled the galvanometric scanner and the 200-ms-long acoustic stimulus. CAM, widefield camera; DM, dichroic mirror; G, diffraction grating; GS, galvanometric scanner; OT, otoscope; EP, earphone.

Fig. 1.

Handheld ISEE system for measuring the acoustic reflex. Near-infrared light from a polarization-maintaining SLD (PM-SLD) was directed through a 50:50 fiber coupler to a reference arm and to the ISEE-otoscope system designed to illuminate the tympanic membrane with the transverse spectrally encoded line. A high-speed spectrometer was used for measuring the interference between the sample and reference arms, and the results were analyzed by a personal computer, which also controlled the galvanometric scanner and the 200-ms-long acoustic stimulus. CAM, widefield camera; DM, dichroic mirror; G, diffraction grating; GS, galvanometric scanner; OT, otoscope; EP, earphone.

Close modal

Computing the oscillatory axial displacement of the membrane from the raw data included [27, 28] dividing each spectral interferogram into 255 windows (32 pixels per window, 50% window overlap), extracting the phase difference between subsequent spectral windows, and computing the instantaneous axial displacement according to Δzi = 4π/λ(xi), where λ denotes the local encoding wavelength at the location xi. For 2D imaging, the spectrally encoded line was slowly scanned across the membrane in the y-axis, allowing at least one full oscillation period to be captured at each y location. The final 2D vibration data Δz(x,y,t) could be either displayed as a short movie of the instantaneous membrane motion or presented as 2D images of the vibration amplitude and phase.

Before measuring the effect of the acoustic reflex, we first tested the expected linear response of the tympanic membrane at different locations. A 40-ms-long acoustic wave with linearly increasing amplitude from zero to 3.8 Pa (105.6 dB) was applied to the left ear of the volunteer, with 1-kHz carrier frequency (Fig. 2, top-left). Four separate line measurements were used to capture the response of four selected regions on the tympanic membrane (Fig. 2, top-right), including the UM, the malleus lateral process, the pars tensa (PT), and the pars flaccida. The measured oscillation traces (blue dots in Fig. 2) were fitted to axial displacement functions having linearly increasing amplitudes Δz (solid red lines in Fig. 2) given by
z=apmaxtTcos2πft+
(1)
where f = 1 kHz denotes the carrier frequency of the applied sound, T = 40 ms denotes the total duration of the wave, pmax = 3.8 Pa denotes the maximum sound pressure at t = 40 ms, and a is the fitted linear coefficient in units of nm/Pa. The four measurements and their corresponding fitted functions, plotted in the four lower panels of Figure 2, show a values between 18.1 nm/Pa and 38 nm/Pa, all with excellent fit qualities (R2 > 0.94).
Fig. 2.

Linear response of the tympanic membrane for stimulus durations shorter than 40 ms. Top-left panel: the recorded linearly increasing sound input at 1-kHz carrier frequency. Top-right: widefield image of the tympanic membrane. UM, umbo; PT, pars tensa; LP, lateral process; PF, pars flaccida. Bottom panels: axial displacements at the corresponding locations (blue markers) and the corresponding linear oscillation fit functions (red lines).

Fig. 2.

Linear response of the tympanic membrane for stimulus durations shorter than 40 ms. Top-left panel: the recorded linearly increasing sound input at 1-kHz carrier frequency. Top-right: widefield image of the tympanic membrane. UM, umbo; PT, pars tensa; LP, lateral process; PF, pars flaccida. Bottom panels: axial displacements at the corresponding locations (blue markers) and the corresponding linear oscillation fit functions (red lines).

Close modal
In order to visualize the effect of the acoustic reflex on the tympanic membrane, we have imaged the membrane vibrations below and above the reflex threshold level, which was measured prior to the experiment to be above 85 dB and below 90 dB using a standard acoustic reflex threshold test at 1-kHz stimulation with a standard tympanometer (Titan, Interacoustics). Selected frames from two eight-frame movies (online suppl. Movie 1; for all online suppl. material, see https://doi.org/10.1159/000538703) are shown in Figure 3 for steady-state acoustic excitations below (Fig. 2a, 84.8 dB) and above (Fig. 2b, 93.3 dB) the measured threshold levels of the reflex. The continuous acoustic stimulus at 1 kHz was applied 1 s before image acquisition (20-kHz line rate, 0.256 s total acquisition time) to allow sufficient time for triggering the reflex. The two oscillation patterns appear quite similar across the frames (Fig. 3a, b), where the effect of the reflex could be identified by comparing the relative difference r between the above-threshold image in Figure 3b and the image expected from a simple linear multiplication of the subthreshold image in Figure 3a:
r%=100z93.3dBpz84.8dBPz84.8dB
(2)
Here Δz93.3dB and Δz84.8dB denote the peak-to-peak oscillation amplitudes at 93.3 dB and 84.8 dB, respectively, and p = 2.655 is the ratio between the applied sound pressure levels. The values of r across the field of view are plotted in Figure 3c, revealing approximately 150% increase in the UM region (marked by small squares in 3a-c), as well as highly nonlinear oscillations (r > 300%) in some of the clear membrane regions, which oscillate at significantly higher amplitudes than those expected from a linear response (r = 100%, gray). The widefield image of the entire region, shown in Figure 3d for reference, also reveals a small abnormal white region at the left-hand side of the image, most likely a myringosclerosis.
Fig. 3.

Imaging the acoustic reflex. a A single frame from an 8-frame ISEE movie of the tympanic membrane at 84.8 dB, 1-kHz excitation. b Same as (a), but at 93.3 dB, above the acoustic reflex threshold. c The normalized difference r (%) between the frame in (b) and the frame in (a) after multiplying the frame in (a) by the difference p in sound pressure (p = 2.655). Gray color (r = 100%) represents linear response. d A widefield image of the measured region on the tympanic membrane. Scale bar represents 1.5 mm.

Fig. 3.

Imaging the acoustic reflex. a A single frame from an 8-frame ISEE movie of the tympanic membrane at 84.8 dB, 1-kHz excitation. b Same as (a), but at 93.3 dB, above the acoustic reflex threshold. c The normalized difference r (%) between the frame in (b) and the frame in (a) after multiplying the frame in (a) by the difference p in sound pressure (p = 2.655). Gray color (r = 100%) represents linear response. d A widefield image of the measured region on the tympanic membrane. Scale bar represents 1.5 mm.

Close modal

For continuously measuring the effect of the acoustic reflex over time, scan-free single-line ISEE [29] was accomplished by positioning the spectrally encoded line (Fig. 4a, dashed line) across the UM (Fig. 4a, square region marked by U) and the nearby clear membrane region (M). A constant-amplitude monotonic acoustic stimulus with 1-kHz carrier frequency was applied to the ear canal during a time window of 200 ms and was recorded using a TRAM TR50 microphone (Fig. 4b). The resulting axial displacement of the UM and membrane regions are shown in Figures 4c and d, respectively, shown next to a similar measurement conducted on a simple rubber membrane as reference (Fig. 4e).

Fig. 4.

Single-line measurement of the umbo and a nearby membrane region show time-dependent amplitude changes. a Widefield image of the measurement region. Dashed line represents the spectrally encoded line, and the umbo and membrane regions are marked by small squares. b Recording of the input acoustic stimulus (constant-amplitude, 200-ms-long, 1-kHz). c Measured umbo (U) oscillations. d Measured membrane (M) oscillations near the umbo. e Same as (c) and (d), but with a rubber membrane instead of the tympanic membrane.

Fig. 4.

Single-line measurement of the umbo and a nearby membrane region show time-dependent amplitude changes. a Widefield image of the measurement region. Dashed line represents the spectrally encoded line, and the umbo and membrane regions are marked by small squares. b Recording of the input acoustic stimulus (constant-amplitude, 200-ms-long, 1-kHz). c Measured umbo (U) oscillations. d Measured membrane (M) oscillations near the umbo. e Same as (c) and (d), but with a rubber membrane instead of the tympanic membrane.

Close modal

Unlike the passive rubber membrane that presents a nearly constant oscillation amplitude (Fig. 4e), the UM region, as well the nearby PT, seem to vary over time with several distinct trends. The UM region (Fig. 4c) responds relatively fast to the initial raise of the acoustic input at t = 0, reaching a local maximum after approximately 5 ms. During the next 5 ms, the oscillations decay slightly and then remain constant at approximately 100 nm amplitude, until t = 40 ms. Between 40 ms and 100 ms, the oscillations seem to be gradually increasing, reaching an amplitude of approximately 140 nm, which then remains constant until 200 ms. At t = 200 ms, the input sound is terminated and the UM oscillations decay rapidly. The PT oscillations generally show similar trends, although the differences appear less obvious (Fig. 4d). As expected, the passive rubber membrane (Fig. 4e), in contrast to the tympanic membrane (Fig. 4c, d), remains almost entirely uniform throughout the entire 200-ms-long acoustic stimulus (Fig. 4e).

In order to better understand the observed oscillation patterns at the UM region, the measured axial displacements in Figure 4c were fitted (Fig. 5) to the well-known equations of a simple damped harmonic oscillator driven by a square-envelope burst monotonic wave at 1-kHz carrier frequency. After a brief period of silence before the stimulus is applied, the measurement window was divided into four time slots: (1) an initial amplitude rise and overshoot (0–40 ms), (2) a gradual buildup of the acoustic reflex effect (40–100 ms), (3) constant oscillations (100–200 ms), and (4) exponential decay after termination of the stimulus (200–220 ms).

Fig. 5.

Fitting the umbo oscillations to a simple model of a damped harmonic oscillator driven by a 200-ms-long 1-kHz acoustic stimulation. The entire measurement (top panel) was divided into four distinct segments: an initial response to the driving oscillations (0–40 ms), activation of the acoustic reflex (40–100 ms), constant-amplitude oscillations (100–200 ms) and oscillation decay (200–220 ms). Blue markers: ISEE measurement. Red curves: fitted harmonic oscillator functions. An excellent correlation (R2 >0.946) was obtained for all parts of the fit.

Fig. 5.

Fitting the umbo oscillations to a simple model of a damped harmonic oscillator driven by a 200-ms-long 1-kHz acoustic stimulation. The entire measurement (top panel) was divided into four distinct segments: an initial response to the driving oscillations (0–40 ms), activation of the acoustic reflex (40–100 ms), constant-amplitude oscillations (100–200 ms) and oscillation decay (200–220 ms). Blue markers: ISEE measurement. Red curves: fitted harmonic oscillator functions. An excellent correlation (R2 >0.946) was obtained for all parts of the fit.

Close modal
Before the sound was applied, no oscillations were measured at the UM, with an RMS noise floor of approximately 6.4 nm. Between zero and 40 ms the measurement was fitted to a damped harmonic oscillator driven by the step-function amplitude from zero to 105.6 dB. The fitting function Δz1(t) was therefore given by
z1t=a1sin2πf1't+1eγ1t+a2cos2πft+2
(3)
where f1′ denotes the oscillation frequency of the UM region, γ1 denotes the damping coefficient, f = 1 kHz is the driving frequency, and the constants a1,2 and ϕ1,2 are the corresponding amplitudes and phases. The results of the curve-fitting algorithm (Matlab) are summarized in Table 1, showing a true oscillation frequency f1′ = 937.6 Hz, and a damping coefficient of γ1 = 192.1 s−1 which is equivalent to approximately 5.2 ms response time.
Table 1.

Resulting fit parameters for harmonic oscillator model

Parameterf1'γ1qa1a2f2'γ2
Description Oscillation frequency Damping coefficient Reflex slope Amplitude Amplitude Oscillation frequency Damping coefficient 
Units Hz 1/s %/ms nm nm Hz 1/s 
Value (fit) 937.6 191.2 0.692 141.9 99.5 926.6 204.2 
Standard error 0.77 3.34 0.014 2.6 0.47 0.49 2.65 
Parameterf1'γ1qa1a2f2'γ2
Description Oscillation frequency Damping coefficient Reflex slope Amplitude Amplitude Oscillation frequency Damping coefficient 
Units Hz 1/s %/ms nm nm Hz 1/s 
Value (fit) 937.6 191.2 0.692 141.9 99.5 926.6 204.2 
Standard error 0.77 3.34 0.014 2.6 0.47 0.49 2.65 
At t = 40 ms, the gradual increase in oscillation amplitudes, which is attributed to the effect of the acoustic reflex, was modeled as a linear increase in amplitude according to
z2t=a21+qtt1cos2πft+3
(4)
where a2 denotes the initial amplitude at t1 = 40 ms, ϕ3 denotes the phase at t = t1, and the parameter q denotes the relative increase in oscillation amplitude during this 60-ms-long period (40–100 ms). The fitting algorithm (R2 = 0.9945) resulted in q = 0.69%/ms, i.e., a 0.69% increase in oscillation amplitude every 1 ms, or a total of 41.52% increase (from 99.5 nm to 140.8 nm) during 60 ms. Between t2 = 100 ms and t3 = 200 ms, the UM oscillations have remained nearly constant, as evident by fitting (R2 = 0.994) the measured trace to a simple constant-amplitude harmonic oscillation at f = 1 kHz, given by
z3t=a3cos2πftt2+4
(5)
Here, a3 = a2(1+qΔt) denotes the constant oscillation amplitude. After the termination of the input sound at t3 = 200 ms, the decaying oscillations were fitted (R2 = 0.973) to a damped harmonic oscillator model with no driving force, given by
z4t=a3cos2πf2'tt3+5eγ2t
(6)
where f2′ and γ2 denote the oscillation frequency and the damping coefficient during the decay, respectively. Interestingly (see Table 1), the parameters f2′ and γ2 during the decay period were significantly (p < 0.0001) different from their equivalents f1 and γ1 during the initial rise and overshoot period (Eq. 3), a difference that could be attributed to the different mechanical conditions exhibited by the tympanic membrane before and after the reflex is activated.

The close match between the model (red curves in Fig. 5) and the ISEE measurement (blue markers) strongly supports the choice of the damped harmonic oscillator model, with excellent correlations (R2 > 0.946) in all of the fitted parameters.

The results presented in Figures 2-5 could be considered equivalent to a standard ipsilateral stimulation test, as the stimulating 1 kHz tone was also served as the “probe” tone that senses the membrane oscillations. Using the relatively strong 93.3 dB sound stimulus allowed us to accurately measure the membrane motion with a high SNR, including the membrane motion before the reflex is activated at approximately 40 ms (middle-left panel in Fig. 5). Clearly, an equivalent contralateral measurement would be extremely valuable to better characterize this effect; however, such contralateral measurement is considerably more challenging because probe tones below certain levels (typically below 90 dB) often result in low oscillation amplitudes that lead to low signal-to-noise ratios.

Such contralateral measurement is shown in Figure 6a, where a subthreshold (84.8 dB) probe tone was applied to the left ear, and an intense over-threshold (93.3 dB) tone was simultaneously applied to the right ear. While the relatively low SNR did not allow us to fit the results to the theoretical harmonic oscillator curves as in Figure 5, the effect of the acoustic reflex was nonetheless still visible by comparing the oscillation amplitude at 20–40 ms, i.e., before the reflex is activated, to the amplitude at 100–200 ms, once the reflex is expected to reach a steady state. The oscillation RMS values where 10.42 ± 0.52 nm and 13.49 ± 0.30 before and after reflex activation, respectively, representing a significant (p < 0.001) 29.5% increase in oscillation amplitude. When using a weaker 84.8 dB stimulating tone in the right ear, which is considered to be below the threshold level of standard tympanometry, the results still showed a significant (p < 0.001) difference between the 20–40 ms and 100–200 ms amplitudes (Fig. 6b), in agreement with the lower reflex thresholds expected from a binaural stimulation [31]. At zero-decibel stimulation at the right ear (Fig. 6c), essentially equivalent to a standard subthreshold ipsilateral measurement, no significant difference was observed between the 20–40 ms and 100–200 ms periods (6c, right-hand panel), indicating that the acoustic reflex was not activated by the probe tone alone.

Fig. 6.

Contralateral ISEE measurement. A subthreshold probe tone of 84.8 dB was applied to the left ear, while an additional stimulation tone was simultaneously applied to the right ear at levels of 93.3 dB (a), 84.8 dB (b), and 0 dB (c). Right-hand panels: bar charts showing the RMS values of the oscillations for the three measurements, comparing the oscillation magnitudes between 20-40 ms and 100–200 ms, i.e., before and after activation of the acoustic reflex.

Fig. 6.

Contralateral ISEE measurement. A subthreshold probe tone of 84.8 dB was applied to the left ear, while an additional stimulation tone was simultaneously applied to the right ear at levels of 93.3 dB (a), 84.8 dB (b), and 0 dB (c). Right-hand panels: bar charts showing the RMS values of the oscillations for the three measurements, comparing the oscillation magnitudes between 20-40 ms and 100–200 ms, i.e., before and after activation of the acoustic reflex.

Close modal

The acoustic reflex is a reaction of the brainstem in response to loud sounds, effectively altering the acoustic transfer function of the middle ear. Standard acoustic reflex threshold measurements conducted with 1-kHz probe frequencies have indicated a significant reduction in the amount of the reflected sound from the middle ear (“reflex deflection” [13, 16]), implying that the acoustic energy is transmitted more effectively into the inner ear. In agreement with these results, the measurements presented in this work show an approximately 42% increase in the UM oscillations at this frequency, representing a notably more efficient energy transfer and hence lower acoustic reflections.

A similar ISEE measurement conducted at 2 kHz stimulation (instead of 1 kHz) above the acoustic reflex threshold, has shown (Fig. 7) that both the UM and the PT were oscillating at relatively constant amplitudes (R2 = 0.971 when fitted to a constant-amplitude harmonic function), not showing any effect of the acoustic reflex. These results agree with previous studies that reported a negligible effect of the acoustic reflex at a similar frequency range [17, 30].

Fig. 7.

Top two panels: the response of the umbo and pars tensa to a 200-ms-long acoustic stimulus at 2 kHz. The stimulus amplitude was 93.3 dB, notably higher than the measured reflex threshold of 85 dB. Lower panel: the corresponding envelopes of the oscillations in the top panels. In contrast to the 1-kHz measurement, here no reflex was detected.

Fig. 7.

Top two panels: the response of the umbo and pars tensa to a 200-ms-long acoustic stimulus at 2 kHz. The stimulus amplitude was 93.3 dB, notably higher than the measured reflex threshold of 85 dB. Lower panel: the corresponding envelopes of the oscillations in the top panels. In contrast to the 1-kHz measurement, here no reflex was detected.

Close modal

Taking advantage of its characteristic high-speed and nanometric sensitivity, ISEE can visualize the effect of the acoustic reflex on the tympanic membrane and compare between different locations on the membrane for studying the temporal progression of the reflex. By applying short, linearly increasing stimuli, we have first verified the linearity of our measurement (Fig. 2). Note that although the increasing sound levels from zero to 105.6 dB cross the measured reflex threshold (which is between 85 and 90 dB), the short 40-ms duration of the measurement is not expected to trigger the reflex, which typically requires response times between 68 and 200 ms [32].

The 2D measurements acquired below and above the reflex threshold (Fig. 3) showed the expected deviation from a linear response; however, the low SNR of the subthreshold image (Fig. 2a) had led to a rather noisy difference image (Fig. 3c). In contrast, the single-line [29] ISEE measurement allowed us to observe the acoustic reflex as it evolves over time, i.e., during activation and until a steady-state oscillation is attained (Fig. 4). The clear results from the UM region, which is the main connection site between the tympanic membrane and the middle ear ossicles, have allowed us to fit the measurement to the simple model of a damped harmonic oscillator driven by a single-frequency force (Fig. 5). Specifically, by comparing the initial amplitude rise at 0–40 ms to the final oscillation decay at 200–220 ms, we have identified significantly (p < 0.00001) different oscillation frequencies (937.6 vs. 926.6 Hz, respectively) and damping coefficients (192.1 vs. 204.2 s−1, respectively). These differences may indicate that the mechanical properties of the middle ear ossicles are significantly altered by the effect of the acoustic reflex.

The system presented in this work has several key advantages over the point measurement conducted by LDV [20, 21], including a high spatial resolution that allows selecting specific regions of interest, and a high SNR provided by spectral-domain interferometry [33, 34], which allows extremely fast measurements that do not require any averaging. Compared to OCT [22, 35, 36], ISEE requires only a single-axis scanning for its 2D-imaging mode and also has a completely scan-free 1D mode that significantly reduces motion artefacts. In a recent paper by Cho et al. [19], both LDV and OCT were used for studying the effect of pulling the stapedius muscle on the velocity and position of the UM region in a human temporal bone (in vitro). At 1 kHz, they have found that the UM velocity was increased with the increasing pulling force, in agreement with our findings on the increased UM oscillation amplitude (Fig. 4c) after activation of the acoustic reflex.

Our current ISEE system has nonetheless several limitations, including a relatively short axial range (typically 1.3 mm) that may limit the field of view at steep regions of the membrane, a noisy 2D-imaging mode that is limited only to high sound stimuli (typically above 90 dB), and a bulky handheld probe (675 g total weight) which could be difficult to operate in the clinical setting. Future versions of the system will utilize custom-build optical and optomechanical components, reducing the overall size and weight and allow high-quality imaging free of motion artefacts. The limited axial range of the system could be improved by using a higher resolution spectrometer and/or by stitching together several measurements, i.e., capturing several images at different reference planes and stitching together the oscillation raw data. Finally, when combined with the system capabilities demonstrated in our previous publications [29, 30], ISEE may offer a full functional examination of the tympanic membrane and the middle ear, revealing both their passive and active response.

In summary, we have imaged the direct effect of the acoustic reflex on the tympanic membrane in a clinically healthy human volunteer. By focusing our measurement on the UM region, we could observe the gradual activation of the reflex, manifested as a 60-ms-long linear increase in oscillation amplitude, approximately 40 ms after the beginning of the stimulus. The measurement was fitted to equations describing a simple damped harmonic oscillator, allowing the extraction of key parameters related to the mechanical properties of the tympanic membrane and the middle ear ossicles. The results may be used to better understand the true mechanical effects of the acoustic reflex and demonstrate the potential of ISEE to noninvasively assess the function of key components in the hearing chain.

The authors wish to thank Rotem Yacoby, Barry Loevsky, and Dr. Amir Wand for their advice on various technical issues encountered in this work.

This study protocol was reviewed and approved by the Rambam Hospital Helsinki Committee, approval protocol RMB-0634-17. Written informed consent was obtained from all participants.

The authors declare no conflicts of interest.

The study was funded in part by the Israel Science Foundation grant (990/19). This work was also supported in part by the Lorry I. Lokey Interdisciplinary Center for Life Sciences and Engineering.

Matan Hamra: methodology, hardware, software, and writing – original draft preparation. Simona Tetin-Schneider: methodology. Sahdi Shinnawi and Mauricio Cohen Vaizer: validation and methodology. Dvir Yelin: supervision, conceptualization, and writing – reviewing and editing.

All data generated or analyzed during this study are included in this article and its online supplementary material files. Further inquiries can be directed to the corresponding author.

1.
Gan
RZ
,
Wood
MW
,
Dormer
KJ
. Dormer, Human middle ear transfer function measured by double laser interferometry system. Otology &amp; neurotology : official publication of the American Otological Society, American Neurotology Society [and] European Academy of Otology and Neurotology;
2004
. Vol.
25
; p.
423
35
.
2.
Goode
RL
,
Killion
M
,
Nakamura
K
,
Nishihara
S
.
New knowledge about the function of the human middle ear: development of an improved analog model
.
Am J Otol
.
1994
;
15
(
2
):
145
54
.
3.
Møller
AR
.
Acoustic reflex in man
.
The J Acoust Soc America
.
1962
;
34
(
9B
):
1524
34
.
4.
Dallos
PJ
.
Dynamics of the acoustic reflex: phenomenological aspects
.
J Acoust Soc America
.
1964
;
36
(
11
):
2175
83
.
5.
Niemeyer
W
.
Relations between the discomfort level and the reflex threshold of the middle ear muscles
.
Audiology
.
1971
;
10
(
3
):
172
6
.
6.
Brask
T
.
The noise protection effect of the stapedius reflex
.
Acta Otolaryngol Suppl
.
1979
;
360
(
Suppl 360
):
116
7
.
7.
de Andrade
KC
,
Camboim
ED
,
Soares
IA
,
Peixoto
MVS
,
Neto
SC
,
Menezes
PL
.
The importance of acoustic reflex for communication
.
Am J Otolaryngol
.
2011
;
32
(
3
):
221
7
.
8.
Jerger
J
,
Anthony
L
,
Jerger
S
,
Mauldin
L
.
Studies in impedance audiometry: III. Middle ear disorders
.
Arch Otolaryngol
.
1974
;
99
(
3
):
165
71
.
9.
Jerger
S
,
Jerger
J
.
Diagnostic value of crossed vs uncrossed acoustic reflexes: eighth nerve and brain stem disorders
.
Arch Otolaryngol
.
1977
;
103
(
8
):
445
53
.
10.
Causon
A
,
Munro
KJ
,
Plack
CJ
,
Prendergast
G
.
The role of the clinically obtained acoustic reflex as a research tool for subclinical hearing pathologies
.
Trends Hear
.
2020
;
24
:
2331216520972860
.
11.
Nuttall
AL
.
Tympanic muscle effects on middle‐ear transfer characteristic
.
J Acoust Soc Am
.
1974
;
56
(
4
):
1239
47
.
12.
Schairer
KS
,
Feeney
MP
,
Sanford
CA
.
Acoustic reflex measurement
.
Ear Hear
.
2013
;
34
(
Suppl 1
):
43s
47s
.
13.
Bennett
MJ
,
Weatherby
LA
.
Multiple probe frequency acoustic reflex measurements
.
Scand Audiol
.
1979
;
8
(
4
):
233
9
.
14.
Keefe
DH
,
Ling
R
,
Bulen
JC
.
Method to measure acoustic impedance and reflection coefficient
.
J Acoust Soc Am
.
1992
;
91
(
1
):
470
85
.
15.
Pyne
JM
,
Lawen
TI
,
Floyd
DD
,
Bance
M
.
The 678 Hz acoustic immittance probe tone: a more definitive indicator of PET than the traditional 226 Hz method
.
J Otolaryngol Head Neck Surg
.
2018
;
47
(
1
):
43
.
16.
Feeney
MP
,
Keefe
DH
.
Acoustic reflex detection using wide-band acoustic reflectance, admittance, and power measurements
.
J Speech Lang Hear Res
.
1999
;
42
(
5
):
1029
41
.
17.
Tonndorf
J
,
Khanna
SM
.
Submicroscopic displacement amplitudes of the tympanic membrane (cat) measured by a laser interferometer
.
J Acoust Soc Am
.
1968
;
44
(
6
):
1546
54
.
18.
Cheng
JT
,
Ghanad
I
,
Remenschneider
A
,
Rosowski
J
.
The onset of nonlinear growth of middle-ear responses to high intensity sounds
.
Hear Res
.
2021
;
405
:
108242
.
19.
Cho
NH
,
Ravicz
ME
,
Puria
S
.
Human middle-ear muscle pulls change tympanic-membrane shape and low-frequency middle-ear transmission magnitudes and delays
.
Hear Res
.
2023
;
430
:
108721
.
20.
Goode
RL
,
Ball
G
,
Nishihara
S
,
Nakamura
K
.
Laser Doppler Vibrometer (LDV): a new clinical tool for the otologist
.
Am J Otol
.
1996
;
17
(
6
):
813
22
.
21.
Jones
HG
,
Greene
NT
,
Ahroon
WA
.
Human middle-ear muscles rarely contract in anticipation of acoustic impulses: implications for hearing risk assessments
.
Hear Res
.
2019
;
378
:
53
62
.
22.
Park
J
,
Cheng
JT
,
Ferguson
D
,
Maguluri
G
,
Chang
EW
,
Clancy
C
, et al
.
Investigation of middle ear anatomy and function with combined video otoscopy-phase sensitive OCT
.
Biomed Opt Express
.
2016
;
7
(
2
):
238
50
.
23.
Wang
RK
,
Nuttall
AL
.
Phase-sensitive optical coherence tomography imaging of the tissue motion within the organ of Corti at a subnanometer scale: a preliminary study
.
J Biomed Opt
.
2010
;
15
(
5
):
056005
.
24.
Cheng
JT
,
Aarnisalo
AA
,
Harrington
E
,
Hernandez-Montes
MDS
,
Furlong
C
,
Merchant
SN
, et al
.
Motion of the surface of the human tympanic membrane measured with stroboscopic holography
.
Hear Res
.
2010
;
263
(
1–2
):
66
77
.
25.
Del Socorro Hernández-Montes
M
,
Furlong
C
,
Rosowski
JJ
,
Hulli
N
,
Harrington
E
,
Cheng
JT
, et al
.
Optoelectronic holographic otoscope for measurement of nano-displacements in tympanic membranes
.
J Biomed Opt
.
2009
;
14
(
3
):
034023
.
26.
Yelin
D
,
Bouma
BE
,
Rosowsky
JJ
,
Tearney
GJ
.
Doppler imaging using spectrally-encoded endoscopy
.
Opt Express
.
2008
;
16
(
19
):
14836
44
.
27.
Grechin
S
,
Yelin
D
.
Imaging acoustic vibrations in an ear model using spectrally encoded interferometry
.
Opt Commun
.
2018
;
407
:
175
80
.
28.
Ilgayev
O
,
Yelin
D
.
Phase-sensitive imaging of tissue acoustic vibrations using spectrally encoded interferometry
.
Opt Express
.
2013
;
21
(
17
):
19681
9
.
29.
Hamra
M
,
Shinnawi
S
,
Vaizer
MC
,
Yelin
D
.
Rapid imaging of tympanic membrane vibrations in humans
.
Biomed Opt Express
.
2020
;
11
(
11
):
6470
9
.
30.
Hamra
M
,
Fridman
L
,
Shinnawi
S
,
Vaizer
MC
,
Yelin
D
.
In vivo optical mapping of the tympanic membrane impulse response
.
Hear Res
.
2023
;
431
:
108723
.
31.
Rawool
VW
,
Parrill
M
.
Objective evaluation of binaural summation through acoustic reflex measures
.
Int J Audiol
.
2018
;
57
(
7
):
493
501
.
32.
Clemis
JD
,
Sarno
CN
.
The acoustic reflex latency test: clinical application
.
Laryngoscope
.
1980
;
90
(
4
):
601
11
.
33.
Yelin
D
,
White
WM
,
Motz
JT
,
Yun
SH
,
Bouma
BE
,
Tearney
GJ
.
Spectral-domain spectrally-encoded endoscopy
.
Opt Express
.
2007
;
15
(
5
):
2432
44
.
34.
Yelin
D
,
Rizvi
I
,
White
WM
,
Motz
JT
,
Hasan
T
,
Bouma
BE
, et al
.
Three-dimensional miniature endoscopy
.
Nature
.
2006
;
443
(
7113
):
765
.
35.
Kirsten
L
,
Schindler
M
,
Morgenstern
J
,
Erkkilä
MT
,
Golde
J
,
Walther
J
, et al
.
Endoscopic optical coherence tomography with wide field-of-view for the morphological and functional assessment of the human tympanic membrane
.
J Biomed Opt
.
2018
;
24
(
3
):
1
11
.
36.
Ramier
A
,
Cheng
JT
,
Ravicz
ME
,
Rosowski
JJ
,
Yun
SH
.
Mapping the phase and amplitude of ossicular chain motion using sound-synchronous optical coherence vibrography
.
Biomed Opt Express
.
2018
;
9
(
11
):
5489
502
.