Hearing Research 384 (2019) 107813

Contents lists available at ScienceDirect

Hearing Research
journal homepage: www.elsevier.com/locate/heares

Research Paper

Recovery from tympanic membrane perforation: Effects on membrane

thickness, auditory thresholds, and middle ear transmission
Lingling Cai a, b, Glenna Stomackin a, Nicholas M. Perez a, d, Xiaohui Lin a,
Timothy T. Jung a, c, Wei Dong a, c, *
a
VA Loma Linda Healthcare System, Loma Linda, CA, 92357, USA
b
Department of Radiology, Shanghai Ninth People’s Hospital, Shanghai JiaoTong University, School of Medicine, Shanghai, China
c
Department of Otolaryngology e Head and Neck Surgery, Loma Linda University Health, Loma Linda, CA, 92350, USA
d
School of Computer Science and Engineering, California State University San Bernardino, San Bernardino, CA, 92407, USA

a r t i c l e i n f o a b s t r a c t

Article history: Sounds delivered to the ear move the tympanic membrane (TM), which drives the middle-ear (ME)
Received 8 July 2019 ossicles and transfers the acoustic energy to the cochlea. Perforations of the TM result in hearing loss
Received in revised form because of less efficient sound conduction through the ME. The patterns of TM motions, and thus ME
20 September 2019
sound transmission, vary with frequency and depend on many factors, including the TM thickness. In this
Accepted 9 October 2019
Available online 15 October 2019
study, we measured TM thickness, auditory brainstem responses (ABR), and ME transmission immedi-
ately following a controlled pars tensa perforation and after 4 weeks of spontaneous recovery in a gerbil
model. It is found that after recovery, the hearing thresholds showed a sloping pattern across fre-
Keywords:
Hearing
quencies: almost back to normal levels at frequencies between 2 and 8 kHz, sloping loss in the low
Middle-ear sound transmission (<2 kHz) and mid-frequency (8e30 kHz) range, and little restoration at frequencies above 30 kHz. This
Optical coherence tomography pattern was confirmed by the measured ME pressure gains. The thickness of the healed TM did not
Tympanic-membrane thickness return to normal but was 2e3 times thicker over a significant portion of the membrane. The increased
Healed tympanic-membrane post- thickness was not limited to the perforated area but expanded into intact regions adjacent to the
perforation perforation, which led to an increased thickness in general. Combined, these results suggest that TM
Middle-ear pressure gain thickness is an important factor in determining its vibration patterns and efficiency to transfer sounds to
the ossicles and thus influencing ME sound transmission, especially for high-frequency sounds. The
results provided both structural and functional observations to explain the conductive hearing loss seen
in patients with abnormal TMs, e.g., caused by otitis media, spontaneously healed post-perforation, or
repaired via tympanoplasty in the clinic.
Published by Elsevier B.V.

1. Introduction transmitted to the malleus that attaches to it, forming the initial
drive to the ossicular chain. The TM moves in a frequency-
The tympanic membrane (TM) has a unique set of properties, dependent manner. At low frequencies the whole TM moves in
such as its elasticity, thickness, and shape that determine how it phase with the stimulus. For higher frequencies, the vibration
vibrates in response to sounds. As the TM vibrates, this motion is pattern becomes more complex with different TM areas moving in
different phases (Cheng et al., 2010, 2013; Khanna and Tonndorf,
1972; Rosowski et al., 2009). The TM thickness is an important
factor in setting these vibration patterns as it directly determines
Abbreviations: 3D, three-dimensional; ABR, auditory brainstem response; CHL, both the elasticity and stiffness of the membrane, and thus impacts
conductive hearing loss; dB, decibel; DPOAE, distortion product otoacoustic emis-
sound transmission. Detailed mapping of TM thickness distribution
sion; IACUC, the Institutional Animal Care and Use Committee; ME, middle ear;
MEPG, middle ear pressure gain; OCT, optical coherence tomography; PTM, pressure is thus important for a more comprehensive understanding of the
at the tympanic membrane near the umbo; PSV, pressure in the scala vestibuli near sound transmission under normal and pathological conditions as it
the stapes; SPL, sound pressure levels; SV, scala vestibuli; TM, tympanic membrane may explain TM related conductive hearing loss (CHL) in the clinic,
* Corresponding author. VA Loma Linda Healthcare System, 11201 Benton Street,
and help the development of theoretical models of middle ear (ME)
Loma Linda, CA, 92357, USA.
E-mail addresses: [email protected], [email protected] (W. Dong). dynamics. However, only limited data exists on TM thickness

https://doi.org/10.1016/j.heares.2019.107813
0378-5955/Published by Elsevier B.V.
2 L. Cai et al. / Hearing Research 384 (2019) 107813

distribution, especially under abnormal conditions. perforation, which equaled a fivefold increase compared to
The thickness of human TM has been measured using conven- normal TMs. Stenfors et al. (1980) performed a similar study in both
tional methods such as optical or electron microscopy with the cats and rats, and found that even 10-weeks post-perforation the
thickness measured over a limited extent (30e150 mm) of the healed perforation area was clearly thicker and did not exhibit a
membrane (e.g. Decraemer and Funnell, 2008; Lim, 1970). Recently, regular structure.
optical coherence tomography (OCT) has been used for mapping In the current study hearing and ME sound transmission was
the distribution of TM thickness. This is a technique that can obtain evaluated immediately after and following a one-month recovery
noninvasive, cross-sectional images from biological tissues using from a 50% perforation of pars tensa in the gerbil ear by measuring
low-coherence interferometry (Huang et al., 1991). It is analogous auditory brainstem responses (ABR) and ME pressure gains
to ultrasound imaging, except that it uses light rather than sound. (MEPG). We also obtained cross-sectional TM images across the
The ability of OCT to provide high-resolution images over a depth of entirety of the membrane using OCT from which we reconstructed
a few millimeters makes it attractive for TM imaging (Ramier et al., TM thickness maps. We found that the perforation in the TM had
2018), and several recent studies have used OCT to visualize the TM closed after four weeks of spontaneous recovery, but that the
and the ME structures behind the TM (Djalilian et al., 2008; Monroy membrane was much thicker (2e3x) compared to pre-perforation.
et al., 2015; Nguyen et al., 2012; Pitris et al., 2001; Shelton et al., Functionally, both the ABR thresholds and MEPG returned to almost
2014; Tan et al., 2018). It has also been used to measure the TM normal for frequencies between 2 and 8 kHz, while responses to
thickness in living humans, yielding estimates between 50 and low and mid-frequency stimuli (<2 kHz and 8e30 kHz) only
120 mm across the membrane (Hubler et al., 2015; Van der Jeught partially recovered. At the highest frequencies (>30 kHz), no signs
et al., 2013). The results obtained by the different methods are of recovery were seen at all. These results indicate that TM thick-
similar, and confirm that the OCT technique could be used for high- ness is an important factor for the coupling of ear-canal sounds to
resolution, real-time imaging of TM and ME structures in the clinic. the inner ear, especially for high-frequency sounds. The current
Gerbils have been widely used in hearing research because they study correlates TM thickness and ME sound transmission in ears
have similar hearing sensitivity/pattern at low frequencies as with normal or healed TMs, which helps to understand the
humans (Ryan, 1976), in contrast to many commonly used labora- mechanism underlying CHL in patients with healed or recon-
tory animals (Heffner and Heffner, 2007). However, the reported structed TM via tympanoplasty, and can be used to formulate and/
normal TM thicknesses in gerbils were different across different or validate models of the ME dynamics.
studies. Von Unge et al.von Unge et al., 1991(reported the gerbil
pars tensa to be 3e5 mm in the central area and 20 mm near the 2. Methods
edges, while Teoh et al. (1997) found the mean pars tensa thickness
to be ~19 mm. It is estimated that conventional histological tech- To establish the relationship between TM thickness and ME
niques, such as tissue fixation and paraffin embedded tissue function, experiments were performed to assess the effects of a
sectioning, can cause ~20e30% decrease in thickness in the central spontaneously healed TM perforation on ME transmission. The left
pars tensa region (Kuypers et al., 2005a). Without using conven- TM of young adult Mongolian gerbils were mechanically perforated
tional histological preparation, Kuypers et al. (2005b) found the by ~50% removal of the pars tensa [(Fig. 1A) and see related details
gerbil pars tensa to be ~6e14 mm, with a mean value of 28 mm along in (Stomackin et al., 2018)], leaving pars flaccida and the ossicular
the annular edge. chain intact. Throughout the text, we refer to this perforation as
TM perforations lead to CHL due to reductions of the effective “TM perforation”. The right ears served as controls. The TM perfo-
drive to the attached ossicle (Stomackin et al., 2018; Voss et al., ration was allowed to spontaneously recover for 4 weeks with
2001). For a moderately sized perforation the TM heals spontane- weekly checks of healing and hearing using otoscopy and pure-tone
ously. However, elevated hearing thresholds remain because of the induced ABR thresholds, respectively. Only animals that remained
thicker and denser healed membrane (Cho et al., 2013; Garth, free of infection and wax build up in the ear canal over the whole
1994). Several studies have explored the thickness of the healed recovery timeline were included in the statistical analysis. After the
TMs in animals. Rahman et al. (2005) reported that the healed area recovery period, ME sound transmission was measured by in vivo
of the pars tensa in rat was about 25 mm thick at two weeks post- recordings of MEPG (see Methods 2.2). Following the data

Fig. 1. Diagram of TM perforation and thickness measurements using the OCT system. (A) Diagram of TM perforation. The perforation was introduced at the inferior part of the pars
tensa to 50% (yellow area) using the umbo as the reference location. (B) 3D scanned OCT image of a normal TM (TM 30ReN). Dotted lines indicate cross sections along which the
thickness was measured. (C) Cross-sectional OCT images of three representative locations as indicated by red dashed lines in panel B. These images were taken from locations above,
at, and under the umbo, perpendicular to the manubrium. The bottom panel is the zoomed in portion of the Slice
4 to illustrate how the thickness of the TM was measured.
 L. Cai et al. / Hearing Research 384 (2019) 107813 3

acquisition, animals were euthanized and both TMs were imme- bursts were recorded at 10 frequencies of 0.5, 2, 4, 5.6, 8, 11.3, 16,
diately harvested and scanned under fresh, and later fixed condi- 22.6, 32, and 45 kHz, at sound pressure levels (SPL) of 0e80 dB SPL
tions using an OCT system to obtain its thickness (see Methods 2.3). in 5 dB steps. To release any ME pressure that may build up during
The latter condition was included to allow a direct comparison of the recordings, the gerbil’s pars flaccida was punctured using a
our results to previous observations that used histological prepa- sterile 100-mm pin before taking ABR measurements. This tiny hole
rations of the TM. is known to only influence hearing at very low frequencies (de La
The care and use of animals were approved by the Institutional Rochefoucauld et al., 2010; Dong et al., 2013) and resealed in two
Animal Care and Use Committee (IACUC) of the VA Loma Linda days. Recording electrodes were placed under the skin with the
Healthcare System. Twenty young adult Mongolian gerbils housed active electrode positioned at the ipsilateral vertex of the skull, the
in autoclaved cages were used for this study, with 11 animals reference electrode placed under the ear canal opening near the
contributed to the post-50%-perforation testing. The other nine bulla location, and a ground electrode placed in the left rear leg.
animals were used to finalize the experimental approach and ABR thresholds were determined from the recordings by visually
explore the spontaneous healing process with different sizes of TM examining the ABR wave-I to determine the lowest sound level at
perforations (Stomackin et al., 2018). which reproducible waveforms were observed.

2.1. Tympanic membrane perforation and monitoring during 2.2. Direct measurements of middle ear pressure gain
healing process
After four weeks of recovery, experiments were performed to
2.1.1. Animal anesthetics directly measure the sound evoked pressures close to the TM (PTM)
Animals were anesthetized by an initial dose of a ketamine and in scala vestibuli (SV) near the stapes (PSV) using a data
(80 mg/kg)/xylazine (10 mg/kg) cocktail i. p., with maintenance acquisition system described in detail elsewhere (Dong et al., 2006,
doses given as needed. While anesthetized, animal body temper- 2019; Stomackin et al., 2018). Middle ear pressure gain is defined
ature was maintained at ~37 using a rectal probe attached to a (Dong and Olson, 2006; Stomackin et al., 2018) as the absolute ratio
Harvard Apparatus heating pad. After the initial TM perforation, as between the SV pressure near the stapes and pressure close to the
well as following the weekly healing checkups and ABR recordings, TM: MEPG ¼ |PSV/PTM|. The sound transmission time through the
an anesthetic reversal dose of 0.2e1 mg/kg atipamezole was given. ME was evaluated by the MEPG phase group delay, defined as t ¼

DF=Df , where F ¼ FPSV e FPTM, where F and f stand for the
2.1.2. Tympanic membrane perforation phase and frequency, respectively.
Before perforating the TM, the animal’s left ear canal was Each animal was deeply anesthetized as described above. For
flushed with a 2% chlorhexidine solution, followed by sterile saline the surgical preparation, the animal’s skull was secured using
to clean. Then, the inferior portion of pars tensa was removed with dental cement (Durelon) to a head holder. The left pinna was
a sterile 30-gauge needle attached to a sterile syringe using the removed and the ispilateral bulla was widely opened. A small hole
manubrium and umbo as reference points (Fig. 1A). Pars tensa and (~200 mm) was created in the bony wall of the SV adjacent to the
perforation area were measured using the OCT system and ImageJ stapes for intracochlear pressure measurements (Dong and Olson,
to confirm that perforation size was ~50% [see also (Stomackin 2006). A micro pressure sensor (~125 mm OD) (Olson and
et al., 2018)]. Great care was taken to use only the minimum Nakajima, 2015) was introduced into SV via this hole and posi-
amount of force necessary to create the desired perforation, while tioned next to the stapes (Dong and Olson, 2006; Olson, 1998). The
keeping the ossicular chain intact. After recovering from the sensor was calibrated in air at both room and body temperature
anesthetic, animals were brought back to autoclaved cages to after construction and calibrated again before and after each
minimize the chance of infection. experiment.
The sound system was coupled to the ear canal opening. It
2.1.3. Healing checkups consisted of a Sokolich ultrasound microphone probe tube that
Over the 4-week recovery time course, the animals were passed through the center of a Y-shaped tube. Each arm of this tube
brought back weekly and anesthetized as described above for was connected to a speaker (Fostex) using plastic tubing. The tip of
otoscopic evaluation and hearing tests. The healing process was the microphone was positioned within ~3 mm of the TM, and
visually monitored using a Stryker video-picture endoscope device acoustic stimuli were calibrated at this point in decibels (dBs) re 20
attached to a Karl Storz 0-degree Hopkins 1218 scope, with which mPap-p. In this configuration, the microphone noise level was ~10 dB
the pictures of the recovering pars tensa were taken. These pictures SPL with a 1-s data acquisition time. The non-perfect sealing of the
helped to determine if there were any other adverse side effects sound system to the ear canal during the measurements influenced
caused by TM perforation, such as wax build up or infection [see responses at frequencies below 1 kHz (de La Rochefoucauld et al.,
images in (Dong et al., 2019)]. Once the perforation had completely 2010).
sealed, the animal’s ear canal was flushed weekly with 1e2 drops of Single-tone stimuli were generated by a commercial system
0.2% salicylic acid solution (Epi-Otic), followed by warm sterile [Tucker-Davis Technologies (TDT) System III] with a sampling rate
saline, to prevent excess wax build up and/or to remove any of 200 kHz driving one Fostex speaker (Dong et al., 2019; Stomackin
existing wax in the ear canal. Gerbils that developed ear infections et al., 2018). Stimulus generation, data acquisition, and analysis
or heavy wax build up at any point during recovery were removed software was written in MATLAB and the TDT visual design studio.
from the study. In the current study, high-level tones of 80 or 90 dB SPL were used
to directly measure pressure responses at the ear canal and in the
2.1.4. Auditory brainstem response (ABR) thresholds SV.
ABR thresholds were determined immediately pre- and post-
perforation, and weekly during the 4-week recovery period for 2.3. Tympanic membrane thickness measurements using optical
the left ear. ABR threshold measurements of the right (control) ears coherence tomography (OCT)
contributed to hearing evaluation under normal conditions.
A detailed description of the ABR recording system is given in an 2.3.1. Tympanic membrane preparations
earlier study (Dong et al., 2019). Briefly, ABRs evoked by single tone After the MEPG measurements, animals were euthanized with
4 L. Cai et al. / Hearing Research 384 (2019) 107813

Pentobarbital (390 mg/ml) and subsequently decapitated. Both the measurements (11 slices & 25e30/slice) were used to create the
normal (right) and healed (left) TMs were harvested by separating corresponding thickness maps using linear interpolation
the TM from connections to the bulla/cochlea, i.e., cutting the ME (MATLAB).
tendons and muscles and disarticulating incus and stapes. The
tympanic ring supporting the TM cone-shape and attached ma- 2.4. Statistical analysis
nubrium were intact as shown in Fig. 1B. By doing this, the TM was
preserved in its natural shape. The fresh specimens were scanned Mean and standard deviation across samples or animals were
using the OCT system immediately after harvest, usually within 30 calculated using MATLAB, i.e., TM thickness measured at similar
min after decapitation (referred to as “fresh condition”). After locations and under similar conditions, hearing evaluations using
scanning, they were placed in a 4% formaldehyde fixative solution ABR or MEPG were averaged at each testing frequency separately.
overnight. Fixed specimens were then scanned using the same OCT
system after 24 h of fixation (termed “fixed condition” in the text). 3. Results

2.3.2. Thickness measurements using OCT 3.1. The thickness of normal tympanic membranes under fresh
The OCT system (Thorlabs, TELESTO Series) and version 4.0 of condition
the Thorlabs ThorImage OCT imaging software allowed us to take
quick 3D scans of the TM within 30 s (Fig. 1B). With this system it is To determine the thickness of normal/non-perforated TM, the
not necessary to flatten the TM, its thickness can be determined ears of nine gerbils were harvested following decapitation. Visually,
across the entirety of its surface without compromising its natural these normal TMs appeared to be smooth and uniform, with all
3D shape. The short scan time minimizes errors caused by tissue areas appearing to be relatively the same (Fig. 1B). Three repre-
preparation and sectioning that were present when using con- sentative cross-sectional OCT images (slices) of two of such normal,
ventional histological techniques. The refractive index was set to fresh TMs are shown in Figs. 1C and 2A. These slices correspond to
1.44, as specified by Van der Jeught et al. (Van der Jeught et al., ~1 mm superior to the umbo, at the umbo, and ~1 mm inferior to
2013). The tympanic ring of each TM was positioned to be as flat the umbo, which we refer to as slice 3, 0 and
4, respectively.
as possible to ensure perpendicular scanning. The umbo was cho- Fig. 2B shows the corresponding thickness measurements at the
sen as the origin of a rectangular coordinate system and the spec- three representative cross-sections. Note that the thickness of the
imen was oriented to have the x-axis perpendicular to the manubrium (which is much thicker than the membrane) was not
manubrium. Positive x- and y-coordinates are in the posterior and measured, causing a gap in the data for slices 0 and 3, respectively.
superior direction, respectively. The scanning pixel sizes used to The TM thickness is not uniform in any of these slices but follows
ensure the entire TM fit within the 3D scanning zone were 4.71, either a W-shaped or a U-shaped profile. That is, the membrane is
4.54, and 2.43 mm along x-, y- and z-axis, respectively. thicker near the outer edges as well as in those regions where the
Once the 3D scan was completed, the cross-sections (slices) umbo/manubrium is attached, ~22e45 mm. This is consistent with
obtained by the “B-scan display” were used to measure the thick- the micrographs in the literature that the fiber layers appear thicker
ness of the TM at various points (Fig. 1C). A total of 11 slices were near the manubrium, umbo, and peripheral rim resulting in thicker
analyzed per TM, with one slice centered at the umbo, 5 slices regions of the TM [e.g., cat, guinea pig: (Lim, 1968); gerbil (Henson
positioned superior to the umbo, and 5 slices positioned inferior to et al., 2005; von Unge et al., 1991)] Also, the TM thickness shows a
the umbo, in ~0.3 mm increments (dotted lines in Fig. 1B and gradual increase from inferior (slice
4) to umbo (slide 0) to su-
Fig. 1C). In each slice, and at approximately every 200 mm, we perior (slice 3) locations within the range of 9e23, 9e26, and
measured the thickness of the TM by manually drawing a line 10e35 mm, respectively. Finally, the anterior side tends to be
segment between its top and bottom surface. These line segments slightly thinner than the posterior side in the TM region superior to
were perpendicular to these surfaces to account for non-horizontal the umbo; inferior to the umbo, no anterior-posterior differences
orientations of the TM (see Fig. 1C bottom panel). These thickness were observed. These trends are not limited to these three slices

Fig. 2. Thicknesses of fresh normal TMs. (A) OCT images of three representative cross-sections of the TM (TM 33ReN, cross-sections of the other example, TM 30ReN, are in
Fig. 1C); (B) Thickness reading of the corresponding cross-sections of normal TMs. Thin lines represent the thickness of each individual TM. Black line and shaded area stand for the
Mean ± SD of 9 TMs. SD: standard deviation. (C) Distribution of the thickness maps of two normal TMs under fresh condition, reconstructed from linear interpretation of 11 cross
section measurements. Black arrows indicate the locations of the representative slices shown in panel B. Dotted lines represent the boundary of the pars tensa.
 L. Cai et al. / Hearing Research 384 (2019) 107813 5

but are observed across the entire TM (Fig. 2C). spontaneously healed TMs were placed in a 4% paraformaldehyde
The distribution of the thickness of each TM was reconstructed fixative solution for 24 h. The fixed TMs were then re-scanned with
from linear interpolation of multiple-point measurements along the OCT system and thickness measurements were repeated. These
the 11 cross-sections and two of them are shown in Fig. 2C. In thickness measurements are presented in Fig. 4 by comparing
general, for the normal TMs, in the region inferior to the umbo, the earlier measurements of the same TMs under fresh (pre-fixing)
thickness reading is smallest and rather constant in the center, conditions. Thus the average number of TMs was different from
which increases when approaching the edge in a range of 9e20 mm. those in Fig. 3. A pairwise comparison (fresh vs. fixed) showed an
In the region superior to the umbo, the thickness distribution is average difference of less than 4 mm at all locations along each of
approximately symmetrical anterior and posterior to the manu- the three representative slices, both for normal and spontaneously
brium, ~15 mm, with thicker readings, ~30 mm, close to the manu- healed conditions. This difference is within the resolution (4.54 mm)
brium and the annulus. One thing to note is that the very top and of the system, indicating that the fixation process did not affect TM
very bottom of the TM were not included in the measurement, thus, thickness. However, our results show a much thicker TM compared
the shape of the TM in our thickness maps falsely portrays a to published data from the same area of fresh gerbil TMs [pink
rounder shape when compared to the actual more ovular shape of regions in Fig. 4A, which are from Fig. 3 in (Kuypers et al., 2005b)].
the TM. The difference might be from the different imaging methodology,
which is further discussed in Discussion section.
3.2. The thickness of spontaneously healed tympanic membranes
under fresh condition 3.4. Hearing and middle ear transmission in animals with
spontaneously healed TMs
Visually, the spontaneously healed TM had a “rougher” and less
uniform appearance than the control ears. The thickness mea- Hearing of the animals was monitored pre-perforation (normal
surements for the freshly imaged, spontaneously healed TM are condition) and weekly following the 50%-TM perforation using ABR
shown in Fig. 3. These data are from six of eight animals; in two thresholds. For gerbils with normal intact TMs, the ABR thresholds
specimens the TM was fused to the tympanic ring in the healed were around 10 dB SPL at 2e22 kHz (black line in Fig. 5A), with
areas, and their thickness measurements were excluded from higher thresholds at lower and higher frequencies. These data are
further analysis. Although the thickness profiles are similarly W- or consistent with previous measures of gerbil hearing (Mason, 2016).
U-shaped, the overall thickness increased. This increase was most The ABR thresholds elevated ~30 dB at all frequencies post perfo-
pronounced in the damaged area (slice
4), where TM thickness ration (dotted blue line in Fig. 5A) and recovered dramatically at
had doubled with a mean thickness of 30e40 mm, but was even frequencies below 20 kHz at week 2 (green line in Fig. 5A), a time at
present in TM regions that were not damaged by the initial perfo- which the perforation was completely sealed. The ABR thresholds
ration (e.g. slices 0 and 3 with thickness of was ~16e43 and continued to recover over the healing process and were almost
25e30 mm, respectively). This can also be appreciated by comparing back to normal levels at frequencies below 16 kHz at week 4 (red
the thickness maps from Figs. 2C and 3C. Pars tensa of the spon- line in Fig. 5A). The ABR thresholds remained elevated at higher
taneously healed TM is markedly thicker, but this thickened area frequencies, i.e., >20 kHz, and appeared to be stable even up to 8
expands to non-damaged TM regions, consistent with observations weeks post-perforation.
at the cellular levels (Johnson et al., 1990). To understand this hearing loss in gerbils with 50% perforated or
healed TMs, MEPG, amplitude and phase responses were directly
3.3. Thickness differences between fresh and fixed conditions measured. The MEPG of gerbils with normal TM was ~20 dB and
almost flat with frequency (black line in Fig. 5B), consistent with
After scanning under fresh conditions, six normal and five results published in the literature [i.e. (Dong and Olson, 2006)]. The

Fig. 3. Thicknesses of fresh spontaneously healed TMs post-50%-perforation. (A) OCT images under fresh conditions of the representative cross-sections of the post-50%-
perforation spontaneously healed TM of animal # 30L-H; (B) Thickness reading of the corresponding cross-sections from individual (thin lines) and averaged TMs (thick red
line). The thick red line and shaded area stand for the Mean ± SD of 6 TMs. As a comparison, the averaged thicknesses of the normal TMs at the same locations were plotted in black
lines (same data as in Fig. 2B). (C) Distribution of the thickness maps of healed TMs post-50%-perforation under fresh condition. Black arrows indicate the locations of the
representative slices shown in panel B. Crosses was used to indicate the original perforation area.
6 L. Cai et al. / Hearing Research 384 (2019) 107813

Fig. 4. Comparisons of thickness of the same TM measured under fresh and fixed conditions. Thickness of the three representative cross-sections of (A) normal and (B) healed
TMs. Black, red and green lines along with shaded area represent Mean ± SD of normal and healed TMs under fresh and fixed conditions, respectively. For comparison, published
data of thicknesses of normal gerbil TMs at the similar locations were also plotted in pink [Fig. 3 in (Kuypers et al., 2005b)].

slight roll-off in MEPG at frequencies less than 1 kHz was due to the resulting in a CHL (Merchant et al., 2003a). TM perforation in pa-
non-perfect sealing of the sound system to the ear canal (de La tients, which tend to heal spontaneously, is commonly caused by
Rochefoucauld et al., 2010). either trauma or infection. However, the healed TMs were found to
Compared to the averaged normal TM conditions, the MEPG be thicker and abnormally dense with elevated hearing thresholds
reduced 20e40 dB post 50% perforation across frequencies (blue (Cho et al., 2013; Garth, 1994). In cases that the TM perforation does
dotted in Fig. 5B), consistent with the increased ABR thresholds. not heal spontaneously, surgery is performed (tympanoplasty) to
After 4 weeks of recovery, the MEPG was back to normal for fre- reconstruct the TM using material such as muscle fascia, peri-
quencies between 2 and 8 kHz (red lines in Fig. 5B). However, at chondrium, and cartilage. Clinical observations indicate that the
lower (<2 kHz) and higher frequencies (>8 kHz), the healed MEPG surgical techniques used to repair a perforated TM can lead to some
did not fully recover. At frequencies of <2 kHz and ~8e25 kHz, the restoration of the CHL postoperatively (Merchant et al., 2003a,
recovered MEPG only deviated by ~5e10 dB from normal, while a 2003b). However, in up to 30% of patients, there remains an
decrease of up to 30e40 dB was present at frequencies above abnormal residual air/bone gap that may vary from 5 to 35 dB
35 kHz. Again, this pattern of loss roughly followed the pattern that (Puria et al., 2013), which continues to produce hearing difficulties.
was seen in ABRs of the spontaneously healed animals. The simi- In our study, we focused on ME sound transmission following
larity between the (changes in) ABR thresholds and MEGPs is more the spontaneous recovery of a perforated pars tensa. The thickness
easily seen in Fig. 5D, in which the relative values (re. normal TM) distributions of gerbil TMs were mapped using a commercial OCT
for these two metrics are directly compared. That is, a change in system. In addition, ME transmission (in terms of MEPG) and ABR
ABR thresholds is mimicked by a change in MEPG (compare solid thresholds of these ears were measured in vivo. We found that,
vs. dashed lines). The non-exact match between the two metrics is although the TM heals (i.e., perforation closes) spontaneously after
at least partially explained by the reduced frequency and intensity perforation, it remains much thicker throughout and normal
resolution used in the determination of the ABR thresholds. This hearing is not restored: there remains a substantial, high-frequency
close correspondence also means that the changes in ABR threshold hearing loss that can be attributed to a less efficient sound trans-
(re. normal) can be used to assess how the sound transmission mission through the middle ear (Fig. 5). Also, the sound trans-
through the ME is affected by the TM pathology. Finally, the phase mission time through the ME appeared to be longer. This is the first
of the MEPG (Fig. 5C) indicate that the ME sound-transmission time that the correlation between TM thickness and hearing
time, evaluated by the phase group delay, t ¼
Df= Df , function has been shown in the same gerbils. In addition, TM
appeared to be longer in ears with healed TM (Fig. 5C), but was thickness distribution maps will be extremely valuable in refining
(nearly) normal in ears with 50% TM perforations. models of TM function, especially under pathological conditions.

4. Discussion 4.1. Determination of the examination time point at four weeks

post-perforation
The structural and physical properties of the TM are important
for transmitting sounds to the ossicular chain, and any variations Otoscopy was used to track the healing process of the TM over
from normal TM conditions likely alter the auditory function, time, which allowed us to ensure that the TMs remained free of
 L. Cai et al. / Hearing Research 384 (2019) 107813 7

infection or excessive wax build up. Characterization of the healing

process is described in detail in a recent publication (Dong et al.,
2019). In general, following a 50%-perforation of the pars tensa,
the gerbil TM began to close quickly, and after one week only ~10%
of the perforation remained. The hole in the TM was completely
sealed by week 2, which is a similar timeline as previously
described for other laboratory rodents (Johnson and Hawke, 1987,
1990; Rahman et al., 2005; Stenfors et al., 1980; Wang et al., 2004).
As examined with an otoscope, little change occurred as the time
course continued after the fourth week, except for a slightly lesser
extent of the thicker-looking opaque tissue on the healed mem-
brane. In addition, ABR thresholds were used to evaluate hearing/
functional restoration over the recovery period. By week 4, low-
frequency ABR thresholds had returned to almost normal levels
(Fig. 5A) at frequencies below 16 kHz. However, there was still
~20 dB threshold elevation at frequencies above 16 kHz, and this
elevation of ABR thresholds remained little change up to 8 weeks.
These observations made us use 4weeks post-perforation as the
terminal measurement point.

4.2. Advantages of using OCT system for thickness measurements

OCT is a revolutionary imaging system that has been adapted to

hearing research quickly. It provides more information that is not
available for the classical essential visualization tools used by an
otolaryngologist. For example, the otoscope and surgical micro-
scope provide only surface imaging within a limited viewing angle,
and without micro-scale features. OCT is a noninvasive, noncontact,
rapid imaging system. It allows surface/3D visualization of tissue
microstructure at high resolution in real time [Fig. 1 and reviews of
(Ramier et al., 2018; Tan et al., 2018)], and has been used to char-
acterize the thickness of TM in human temporal bones, in human
subjects with normal TMs as well as in otitis media patient with
abnormal TMs (Cho et al., 2015; Dsouza et al., 2018; Hubler et al.,
2015; Park et al., 2018; Van der Jeught et al., 2013). In our study,
we have demonstrated that a commercially available system can be
used to map gerbil TM thickness distribution in a more efficient
way compared to other conventional methods (e.g. histology). TM
thickness measurements were obtained by taking a 3D scan of an
entire TM within several seconds. The 3D scanning image was then
easily sliced into many 2D cross-section images of the TM using the
function of “B-Scans” (Fig. 1), and TM thickness measurements
were then performed at each cross-section with a resolution of
3e5 mm. Our results confirm the potential application of OCT in
both the clinic and in research laboratories as a diagnostic aid for
otologic conditions.

4.3. TM thickness distribution under intact condition

In a normal gerbil TM, we found that the TM thickness is not

uniform. The central portion of the gerbil pars tensa between the
annulus and the manubrium was the thinnest, ~13e16 mm on
average. The thickness increased close to the bone edge,
~20e30 mm in general (Figs. 2 and 4), resulting in a W- or U-shaped
thickness pattern, like the one seen by Kuypers et al. (2005a). The
thickness measured in our study, however, appeared to be thicker

4PTM; (D) Changes in ABR threshold and MEPG relative to normal conditions. Solid and
dashed lines stand for ABR threshold and MEPG, respectively. For comparison,
mean±SD across animals with intact, immediate post-50%-perforation, post-perfora-
tion week 2 and week 4 defined as healed TMs were plotted as black, dotted-blue,
green and red lines and shaded areas. Averaged data of post-perforation was from
another group of animals for study of effects of size-dependent TM perforation on
Fig. 5. Hearing evaluation of animals with spontaneously healed TMs. (A) ABR
middle ear transmissions (Stomackin et al., 2018).
thresholds; (B) MEPG defined as |PSV/PTM|; (C) MEPG phase responses defined as 4PSV -
8 L. Cai et al. / Hearing Research 384 (2019) 107813

than theirs at similar locations as our example slices (see Fig. 4). The sound transmission through the ME. For spontaneously healed
difference here might be from the preparations of the TMs. In our TMs, the variations include increased thickness and amount of fi-
study, the TMs were scanned within 30 min post mortem without bers exhibited with an obvious lacking of the reorganization of the
any further processing, referred to as the “fresh” condition, and fiber layer. These structural changes suggest a variation in stiffness
again 24 h after fixation (“fixed” condition). We found no difference and increased mass of the healed TM, properties that will affect the
in the thickness estimates between these two methods (Fig. 4). In mechanical responses of this membrane.
Kuypers’s study, the TMs were stained by soaking for 30 min in Effects of mechanical properties of the TM on ME sound trans-
water-based Van Giesson dye, which they claim does not cause mission has been detailed described in theoretical models, i.e., 3D
dehydration or shrinking (Kuypers et al., 2005a, 2005b). If not the finite-element (FE) models (Caminos et al., 2018; O’Connor et al.,
fixing process, the difference may arise from the mounting of the 2017). The mass of the TM was found to have little effect on the
TMs. They mounted the TMs on a standard cover glass with their MEPG at low frequencies [Fig. 3 in (O’Connor et al., 2017)]; however,
medial sides downwards, facing the objective lens. Limited by the it showed an inverse relationship to the MEPG at mid-to-high fre-
working distance of their confocal lens (230 mm), the membranes quencies. According to the 3D FE model prediction, a 5e10x in-
had to be flattened, from their natural cone shape, across the glass. crease in mass caused reduction in the MEPG amplitude up to 20 dB
Although TM folds were avoided in the regions that had been in the mid- and high-frequency region. In addition, increased mass
measured, it is not clear whether these manipulations decreased led to up to 2x longer ME transmission times. In their model, the
the thickness of the membranes. influence of TM fiber distribution/stiffness in radial, circumferen-
Other groups have measured the thickness of the gerbil TM. tial, and transverse direction was also discussed. Their results
However, the majority of these studies used histological sectioning. depend on several model restrictions, e.g., isotropic vs orthotropic
It is not clear if, and to what extent, the histological preparation TM, “soft” vs “hard” boundary conditions, variation in shear vs
techniques affect the TM thickness. Using histological approaches, Young’s modulus, but indicate that variation in the stiffness of
the gerbil TM was found to be anywhere from 3 to 20 mm in general radial collagen fibers has the largest effect on the MEPG, especially
(Teoh et al., 1997; von Unge et al., 1991; von Unge et al., 1999). Our at high frequencies. These predictions are consistent with mea-
study narrowed the range down to ~13e16 mm in the central region surements from human cadaveric temporal bones in that radial
between the manubrium and annulus, and ~20e30 mm along the distributed fibers played greater role in ME sound transmission at
annular edge. frequencies above 4 kHz than circumferential arranged fibers
(O’Connor et al., 2008). In another 3D FE prediction, effects of TM
4.4. TM thickness distribution after spontaneously healing thickness were discussed (Caminos et al., 2018). Here, an increased
TM thickness was considered to increase both the mass and stiff-
As the perforation starts to heal, the edges and the areas just ness. Model results showed an overall reduction pattern across
past the edges start to thicken and advance to close the hole in a frequencies in the ME transmission but more so in the low fre-
non-symmetrical way, followed by a migration of spurs of epithelial quency region.
cells towards the defects (Johnson et al., 1990). Granulation tissue In the current study, the ME transmission evaluated by ABR and
progressively invades the area with abundant collagen fibers. A directly measured using MEPG in ears with healed TMs appeared to
“leading edge” forms that looks both rough and jagged and mi- be back to normal at frequencies below 8 kHz, less efficient (<10 dB
grates to close the perforation. Even after complete closure of the loss) at low- (<2 kHz) and mid-frequencies (8e25 kHz) and 20 dB
perforation, the thickness of both the healed areas and the areas or more loss at higher frequencies (Fig. 5). These results are
near the perforation are non-uniform and thicker than normal consistent with findings of hearing restoration at 2 and 4 kHz in
(Johnson et al., 1990; Rahman et al., 2005; Stenfors et al., 1980). patients withspontaneously healed TM perforations who experi-
Thus, TMs display a thickened membrane not only over the healed- enced blast injuries during the Boston Marathon on April 15, 2013
perforation locus, but also on areas adjacent to the perforation (Remenschneider et al., 2014). Due to the extended frequency range
(Rahman et al., 2005, 2007). As described in details in an earlier of gerbil hearing, their 16 kHz corresponds to 4 kHz in humans. In
report (Dong et al., 2019), by week-4 the gerbil TM-50%-perforation addition, we found that the delay of ME sound transmission
appeared to be completely healed and some of the visible scar (evaluated by the phase group delay) appeared to be longer after
tissue was starting to be reduced. However, the thickness mea- TM healing. According to the model predictions discussed above,
surements of the healed TM revealed a much thicker pars tensa the variations in the ME transmission (amplitude and phase) we
than a normal TM, especially around the perforation area. Also, the observed in these healed ears appeared to result from the com-
thickness was not homogenous across the healed area (Figs. 3 and bined effects of an increased mass and a non-organized fiber
4). The central portion of the TM between the annulus and ma- distribution.
nubrium measured anywhere from 25 to 65 mm, with the edges Our results also offer some explanations for the outcomes
again slightly thicker. Our results are consistent with reports in the following tympanoplasty. This procedure re-establishes the barrier
literature, e.g., two-weeks post-perforation, rat TM appear to be 5 between the ear canal and ME when a TM perforation does not heal
times thicker than normal (Rahman et al., 2005) and even after 10 spontaneously. For the reconstruction, muscle fascia and cartilage
weeks, or half a year, the healed perforation area was clearly thicker are routinely used Mouna (2019). The cartilage plate is normally
in cats and rats (Rahman et al., 2007; Stenfors et al., 1980). The prepared with a thickness of ~500 mm (i.e. Beutner et al., 2010)
thicker TM was due to the irregular structure exhibited in these significantly thicker than a normal human TM (50e120 mm)
animals, i.e., ingrowth of fibroblasts and extracellular matrix with (Hubler et al., 2015; Van der Jeught et al., 2013). And the temporalis
fiber-like structures without an obvious system of orientation fascia was demonstrated to not significantly remodel, change
(Henson et al., 2005; Lim, 1968; von Unge et al., 1991). Thus, the thickness or change fiber structure following successful tympano-
thickness measurements help to confirm the deviations from plasty (Trakimas et al., 2018); its thickness was about 2x that of the
normal in functional evaluations in the healing process. normal TM even sixteen years after the reconstruction. The
reconstructed, and thickened, TM effectively restored low-
4.5. The role of TM in sound transmission and clinical significance frequency hearing, but did little to recover hearing at high fre-
quencies. Thus, as suggested by our results, the mechanical prop-
Variations in the physical properties of the TM directly influence erties of the TM are important to sound transmission, especially at
 L. Cai et al. / Hearing Research 384 (2019) 107813 9

the high-frequency region. To improve hearing following tympa- Tympanic Membrane, Chronic Otitis Media. Pathogenesis-Oriented Theropeutic
Management. Kugler Publications, The Hague, Amsterdam, The Netherlands,
noplasty it seems pertinent to mimic the structural properties of
pp. 51e84.
the TM, rather than simply “closing the gap”, perhaps using ad- Djalilian, H.R., Ridgway, J., Tam, M., Sepehr, A., Chen, Z., Wong, B.J., 2008. Imaging
vances in 3D printing and stem cell techniques (Kozin et al., 2016). the human tympanic membrane using optical coherence tomography in vivo.
Otol. Neurotol. 29, 1091e1094.
Dong, W., Olson, E.S., 2006. Middle ear forward and reverse transmission in gerbil.
In summary J. Neurophysiol. 95, 2951e2961.
Dong, W., Varavva, P., Olson, E.S., 2013. Sound transmission along the ossicular
chain in common wild-type laboratory mice. Hear. Res. 301, 27e34.
Mapping the TM thickness is valuable in understanding the ME Dong, W., Stomackin, G., Lin, X., Martin, G.K., Jung, T.T., 2019 Jul. Distortion product
sound transmission under normal and pathological conditions. We otoacoustic emissions: sensitive measures of tympanic -membrane perforation
used a gerbil model with a spontaneously healed TM following a 4- and healing processes in a gerbil model. Hear. Res. 378, 3e12.
Dsouza, R., Won, J., Monroy, G.L., Hill, M.C., Porter, R.G., Novak, M.A., Boppart, S.A.,
week recovery post-perforation of pars tensa to map the TM
2018. In vivo detection of nanometer-scale structural changes of the human
thickness using a high-resolution commercial OCT imaging system. tympanic membrane in otitis media. Sci. Rep. 8, 8777.
The system allowed us to reconstruct the TM thickness distribution Garth, R.J., 1994. Blast injury of the auditory system: a review of the mechanisms
in an efficient way. The normal TMs were not perfectly homoge- and pathology. J. Laryngol. Otol. 108, 925e929.
Heffner, H.E., Heffner, R.S., 2007. Hearing ranges of laboratory animals. J. Am. Assoc.
neous, with the thinner reading of ~13e16 mm on average in the Lab. Anim. Sci. 46, 20e22.
central portion of the pars tensa, and gradually increased thickness Henson, M.M., Madden, V.J., Rask-Andersen, H., Henson Jr., O.W., 2005. Smooth
while approaching the annulus, ~20e30 mm thick in general. The muscle in the annulus fibrosus of the tympanic membrane in bats, rodents,
insectivores, and humans. Hear. Res. 200, 29e37.
healed TMs showed rough surface, displayed thickened membrane, Huang, D., Swanson, E.A., Lin, C.P., Schuman, J.S., Stinson, W.G., Chang, W., Hee, M.R.,
~2e3 times to the normal, not only over the healed-perforation Flotte, T., Gregory, K., Puliafito, C.A., et al., 1991. Optical coherence tomography.
locus, but also on areas adjacent to the perforation. The thickness Science 254, 1178e1181.
Hubler, Z., Shemonski, N.D., Shelton, R.L., Monroy, G.L., Nolan, R.M., Boppart, S.A.,
distribution demonstrated the variation of physical properties in 2015. Real-time automated thickness measurement of the in vivo human
the healed TM , and correlates to the non-recovered ME sound tympanic membrane using optical coherence tomography. Quant. Imaging Med.
transmission, especially at high frequencies, which is consistent Surg. 5, 69e77.
Johnson, A., Hawke, M., 1987. The function of migratory epidermis in the healing of
with model predictions. tympanic membrane perforations in Guinea-pig. A photographic study. Acta
Otolaryngol. 103, 81e86.
Funding Johnson, A.P., Smallman, L.A., Kent, S.E., 1990. The mechanism of healing of tym-
panic membrane perforations. A two-dimensional histological study in Guinea
pigs. Acta Otolaryngol. 109, 406e415.
The research work was performed at facilities provided by the Khanna, S.M., Tonndorf, J., 1972. Tympanic membrane vibrations in cats studied by
Department of Veterans Affairs. This study was supported by VA time-averaged holography. J. Acoust. Soc. Am. 51, 1904e1920.
Kozin, E.D., Black, N.L., Cheng, J.T., Cotler, M.J., McKenna, M.J., Lee, D.J., Lewis, J.A.,
Rehabilitation Research and Development Merit Award C2296-R Rosowski, J.J., Remenschneider, A.K., 2016. Design, fabrication, and in vitro
(Dong), The National Institute on Deafness and Other Communi- testing of novel three-dimensionally printed tympanic membrane grafts. Hear.
cation Disorders (NIDCD) R01DC011506 (Dong), and the Depart- Res. 340, 191e203.
Kuypers, L.C., Decraemer, W.F., Dirckx, J.J., Timmermans, J.P., 2005a. Thickness dis-
ment of Otolaryngology Head & Neck Surgery, at Loma Linda
tribution of fresh eardrums of cat obtained with confocal microscopy. JARO J.
University Health. These contents do not represent the views of the Assoc. Res. Otolaryngol. : JARO 6, 223e233.
U.S. Department of Veterans Affairs or the United States Kuypers, L.C., Dirckx, J.J., Decraemer, W.F., Timmermans, J.P., 2005b. Thickness of the
gerbil tympanic membrane measured with confocal microscopy. Hear. Res. 209,
Government.
42e52.
Lim, D.J., 1968. Tympanic membrane. Electron microscopic observation. I: pars
Acknowledgments tensa. Acta Otolaryngol. 66, 181e198.
Lim, D.J., 1970. Human tympanic membrane. An ultrastructural observation. Acta
Otolaryngol. 70, 176e186.
The authors would like to thank Justin Li and Brandon Shin to Mason, M.J., 2016. Structure and function of the mammalian middle ear. II: inferring
work out the method of using OCT scanning to perform TM thick- function from structure. J. Anat. 228, 300e312.
Merchant, S.N., Rosowski, J.J., McKenna, M.J., 2003a. Tympanoplasty. Operative
ness measurements. And thanks to Sebastiaan W. F. Meenderink for
Techiniques in Otolaryngology-Head and Neck Surgery, vol.14, pp. 224e236.
the helpful comments, discussions, and modification on the Merchant, S.N., McKenna, M.J., Mehta, R.P., Ravicz, M.E., Rosowski, J.J., 2003b.
manuscript. Middle ear mechanics of Type III tympanoplasty (stapes columella): II. Clinical
studies. Otol. Neurotol. 24, 186e194.
Monroy, G.L., Shelton, R.L., Nolan, R.M., Nguyen, C.T., Novak, M.A., Hill, M.C.,
References McCormick, D.T., Boppart, S.A., 2015. Noninvasive depth-resolved optical mea-
surements of the tympanic membrane and middle ear for differentiating otitis
Beutner, D., Huttenbrink, K.B., Stumpf, R., Beleites, T., Zahnert, T., Luers, J.C., media. The Laryngoscope 125, E276eE282.
Helmstaedter, V., 2010. Cartilage plate tympanoplasty. Otol. Neurotol. 31, Mouna, B., Khalifa M, Ghammem M, Limam M, Meherzi A, Kermani W, Abdelkefi M.,
105e110. 2019. Cartilage and Fascia Graft In Type 1 Tympanoplasty: Comparison of
Caminos, L., Garcia-Manrique, J., Lima-Rodriguez, A., Gonzalez-Herrera, A., 2018. Anatomical and Audological Results. J Craniofac Surg. 30 (4), e297ee300.
Analysis of the mechanical properties of the human tympanic membrane and https://doi.org/10.1097/SCS.0000000000005278. PMID: 30839461.
its influence on the dynamic behaviour of the human hearing system. Appl. Nguyen, C.T., Jung, W., Kim, J., Chaney, E.J., Novak, M., Stewart, C.N., Boppart, S.A.,
Bionics Biomechanics 2018, 1736957. 2012. Noninvasive in vivo optical detection of biofilm in the human middle ear.
Cheng, J.T., Hamade, M., Merchant, S.N., Rosowski, J.J., Harrington, E., Furlong, C., Proc. Natl. Acad. Sci. U. S. A. 109, 9529e9534.
2013. Wave motion on the surface of the human tympanic membrane: holo- O’Connor, K.N., Cai, H., Puria, S., 2017. The effects of varying tympanic-membrane
graphic measurement and modeling analysis. J. Acoust. Soc. Am. 133, 918e937. material properties on human middle-ear sound transmission in a three-
Cheng, J.T., Aarnisalo, A.A., Harrington, E., Hernandez-Montes Mdel, S., Furlong, C., dimensional finite-element model. J. Acoust. Soc. Am. 142, 2836.
Merchant, S.N., Rosowski, J.J., 2010. Motion of the surface of the human tym- O’Connor, K.N., Tam, M., Blevins, N.H., Puria, S., 2008. Tympanic membrane collagen
panic membrane measured with stroboscopic holography. Hear. Res. 263, fibers: a key to high-frequency sound conduction. The Laryngoscope 118,
66e77. 483e490.
Cho, N.H., Lee, S.H., Jung, W., Jang, J.H., Kim, J., 2015. Optical coherence tomography Olson, E.S., 1998. Observing middle and inner ear mechanics with novel intra-
for the diagnosis and evaluation of human otitis media. J. Korean Med. Sci. 30, cochlear pressure sensors. J. Acoust. Soc. Am. 103, 3445e3463.
328e335. Olson, E.S., Nakajima, H.H., 2015. A Family of Fiber-Optic Based Pressure Sensors for
Cho, S.I., Gao, S.S., Xia, A., Wang, R., Salles, F.T., Raphael, P.D., Abaya, H., Wachtel, J., Intracochlear Measurements. SPIE BIOS.
Baek, J., Jacobs, D., Rasband, M.N., Oghalai, J.S., 2013. Mechanisms of hearing Park, K., Cho, N.H., Jeon, M., Lee, S.H., Jang, J.H., Boppart, S.A., Jung, W., Kim, J., 2018.
loss after blast injury to the ear. PLoS One 8, e67618. Optical assessment of the in vivo tympanic membrane status using a handheld
de La Rochefoucauld, O., Kachroo, P., Olson, E.S., 2010. Ossicular motion related to optical coherence tomography-based otoscope. Acta Otolaryngol. 138, 367e374.
middle ear transmission delay in gerbil. Hear. Res. 270, 158e172. Pitris, C., Saunders, K.T., Fujimoto, J.G., Brezinski, M.E., 2001. High-resolution im-
Decraemer, W.F., Funnell, W.R., 2008. Anatomical and Mechanical Properties of the aging of the middle ear with optical coherence tomography: a feasibility study.
10 L. Cai et al. / Hearing Research 384 (2019) 107813

Arch. Otolaryngol. Head Neck Surg. 127, 637e642. membrane perforation on middle ear transmission in gerbil. Hear. Res. 373,
Puria, S., Fay, R.R., Popper, A., 2013. The Middle Ear, Springer Handbook of Auditory 48e58.
Research, vol.46. Springer-Verlag, New York, pp. 1e308, 1. Tan, H.E.I., Santa Maria, P.L., Wijesinghe, P., Francis Kennedy, B., Allardyce, B.J.,
Rahman, A., Hultcrantz, M., Dirckx, J., von Unge, M., 2007. Structural and functional Eikelboom, R.H., Atlas, M.D., Dilley, R.J., 2018. Optical coherence tomography of
properties of the healed tympanic membrane: a long-term follow-up after laser the tympanic membrane and middle ear: a review. Otolaryngol. Head Neck
myringotomy. Otol. Neurotol. 28, 685e691. Surg. 159, 424e438.
Rahman, A., Hultcrantz, M., Dirckx, J., Margolin, G., von Unge, M., 2005. Fresh Teoh, S.W., Flandermeyer, D.T., Rosowski, J.J., 1997. Effects of pars flaccida on sound
tympanic membrane perforations heal without significant loss of strength. Otol. conduction in ears of Mongolian gerbil: acoustic and anatomical measure-
Neurotol. 26, 1100e1106. ments. Hear. Res. 106, 39e65.
Ramier, A., Cheng, J.T., Ravicz, M.E., Rosowski, J.J., Yun, S.H., 2018. Mapping the phase Trakimas, D.R., Ishai, R., Ghanad, I., Black, N.L., Kozin, E.D., Cheng, J.T.,
and amplitude of ossicular chain motion using sound-synchronous optical Remenschneider, A.K., 2018. Otopathologic evaluation of temporalis fascia
coherence vibrography. Biomed. Opt. Express 9, 5489e5502. grafts following successful tympanoplasty in humans. The Laryngoscope 128,
Remenschneider, A.K., Lookabaugh, S., Aliphas, A., Brodsky, J.R., Devaiah, A.K., E351eE358.
Dagher, W., Grundfast, K.M., Heman-Ackah, S.E., Rubin, S., Sillman, J., Tsai, A.C., Van der Jeught, S., Dirckx, J.J., Aerts, J.R., Bradu, A., Podoleanu, A.G., Buytaert, J.A.,
Vecchiotti, M., Kujawa, S.G., Lee, D.J., Quesnel, A.M., 2014. Otologic outcomes 2013. Full-field thickness distribution of human tympanic membrane obtained
after blast injury: the Boston Marathon experience. Otol. Neurotol. 35, with optical coherence tomography. JARO J. Assoc. Res. Otolaryngol. : JARO 14,
1825e1834. 483e494.
Rosowski, J.J., Cheng, J.T., Ravicz, M.E., Hulli, N., Hernandez-Montes, M., von Unge, M., Bagger-Sjoback, D., Borg, E., 1991. Mechanoacoustic properties of the
Harrington, E., Furlong, C., 2009. Computer-assisted time-averaged holograms tympanic membrane: a study on isolated Mongolian gerbil temporal bones. Am.
of the motion of the surface of the mammalian tympanic membrane with J. Otol. 12, 407e419.
sound stimuli of 0.4-25 kHz. Hear. Res. 253, 83e96. von Unge, M., Decraemer, W.F., Dirckx, J.J., Bagger-Sjoback, D., 1999. Tympanic
Ryan, A., 1976. Hearing sensitivity of the Mongolian gerbil, Meriones unguiculatis. membrane displacement patterns in experimental cholesteatoma. Hear. Res.
J. Acoust. Soc. Am. 59, 1222e1226. 128, 1e15.
Shelton, R.L., Jung, W., Sayegh, S.I., McCormick, D.T., Kim, J., Boppart, S.A., 2014. Voss, S.E., Rosowski, J.J., Merchant, S.N., Peake, W.T., 2001. Middle-ear function with
Optical coherence tomography for advanced screening in the primary care of- tympanic-membrane perforations. I. Measurements and mechanisms. J. Acoust.
fice. J. Biophot. 7, 525e533. Soc. Am. 110, 1432e1444.
Stenfors, L.E., Carlsoo, B., Salen, B., Winblad, B., 1980. Repair of experimental tym- Wang, W.Q., Wang, Z.M., Chi, F.L., 2004. Spontaneous healing of various tympanic
panic membrane perforations. Acta Otolaryngol. 90, 332e341. membrane perforations in the rat. Acta Otolaryngol. 124, 1141e1144.
Stomackin, G., Kidd, S., Jung, T.T., Martin, G.K., Dong, W., 2018. Effects of tympanic