Static and dynamic forces in the incudostapedial joint gap

Hearing Research xxx (xxxx) xxx

Contents lists available at ScienceDirect

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

Research Paper

Static and dynamic forces in the incudostapedial joint gap Martin Koch a, *, Till Moritz Eßinger a, Martin Angerer b, Thomas Stoppe a, Matthias Bornitz a, Marcus Neudert a, Thomas Zahnert a a

Technische Universitaet Dresden, Faculty of Medicine Carl Gustav Carus, Department of Otorhinolaryngology, ERCD Ear Research Center Dresden, Germany b MED-EL Medical Electronics, Innsbruck, Austria

a r t i c l e i n f o

a b s t r a c t

Article history: Received 31 July 2018 Received in revised form 9 November 2018 Accepted 12 February 2019 Available online xxx

Dynamic pressure at the tympanic membrane is transformed and subsequently transferred through the ossicular chain in the form of forces and moments. The forces are primarily transferred to the inner ear. They are transferred partly to the stapedial annular ligament which exhibits non-linear behavior and stiffens for larger static forces. In unventilated middle ears, static pressure is additionally transferred to the ossicles. The purpose of this study was to measure the force inside the ossicular chain as a physiological parameter. We determined the forces which act for dynamic sound transmission and for static load on the ossicular chain. The study is the first one which introduces these forces. The static forces have direct impact on clinically relevant questions for middle ear reconstructions with passive or active prosthesis. The dynamic forces have an impact on the development of middle ear sensors. Quasi-static forces in the incudostapedial joint (ISJ) gap were measured with two different sensor types in 17 temporal bones. The sensing elements, a single crystal piezo and a strain gauge element for validation, were bonded to a thin flexible titanium plate and encapsulated in a titanium housing to allow the acquisition of the applied force signal inside the ossicular chain. Dynamic forces were measured in 11 temporal bones with the piezo sensor. We measured a static force of 23 mN in the ISJ after sensor insertion. The mean force for dynamic physiological acoustic excitation from 250 Hz to 6 kHz was 26 mN/Pa. If the tympanic membrane is loaded with a static pressure, the static force in the ISJ increases up to 1 N for a maximum static pressure load scenario of 30 kPa. © 2019 Elsevier B.V. All rights reserved.

Keywords: Incudostapedial joint gap Force measurement Piezoelectric sensor Strain gauge sensor

1. Introduction 1.1. Middle ear pressure and pressure equilibration The human ear as a sensory organ can essentially be considered as a dynamic pressure sensor. The effective range for dynamic pressure (i.e. sound pressure) in a healthy ear is 2.2 mPae6.3 Pa, or 1e110 dB sound pressure level (dB SPL) at the ear canal between perceptional threshold and level of discomfort for the audiological mid frequency range between 1 kHz and 4 kHz (Lehnhardt, 2009).

Abbreviations: AL, annular ligament; AMEI, active middle ear implant; ASTM, American Society for Testing and Materials; dB SPL, decibels sound pressure level; CI, confidence interval; FE, finite elements; FEM, finite elements model; ISJ, incudostapedial joint; LDV, Laser Doppler Vibrometer; METF, Middle ear transfer function; OC, ossicular chain; SG, strain gauge; TB, temporal bone; TI, tolerance interval; TM, tympanic membrane * Corresponding author. E-mail address: [email protected] (M. Koch).

The atmospheric pressure itself has a much higher quasi-static pressure (110 kPa at sea level (Klose, 2008),). Normally, this does not strain the tympanic membrane (TM) and the ossicular chain (including the stapedial annular ligament (AL), because the middle ear is equipped with a mechanism for pressure equalization (Feldmann, 1973). The most important mechanism for this is the opening of the Eustachian tube, which starts at a quasi-static overpressure of 2 kPa in the middle ear (Samuels, 2004) (Mirza, 2005). Samuels states that for a negative pressure (overpressure inside the mastoid cavity), especially if applied rapidly, the pressure equilibration can fail due to causes such as rhinitis and methods like swallowing or Valsalva's maneuver may also fail. The resultant pressure at the TM translates into a force on the ossicular chain (OC) and the AL. Beside a temporary decline in stapes footplate movement as described by (Vorwerk et al., 1999), the pressure difference can lead to otic barotrauma.

https://doi.org/10.1016/j.heares.2019.02.004 0378-5955/© 2019 Elsevier B.V. All rights reserved.

Please cite this article as: Koch, M et al., Static and dynamic forces in the incudostapedial joint gap, Hearing Research, https://doi.org/10.1016/ j.heares.2019.02.004

2

M. Koch et al. / Hearing Research xxx (xxxx) xxx

Table 1 Load scenarios of quasi-static pressures on the TM without ear clearing (except Valsalva which leads to ear clearing). Load scenario

Pressure difference at TM

Literature

Regular pressure variations, such as air pressure variations due to weather conditions or small altitude changes Valsalva's maneuver 2 m diving Aircraft landing 3 m diving

1 kPa

Klose (2008)

5 kPa 20 kPa 25 kPa 30 kPa

(Looga, 2005) Calculated from p ¼ r  g  h Samuels (2004) Calculated from p ¼ r  g  h

1.2. Ossicular chain force measurement The effects of quasi-static pressure have been studied for several topics, such as TM movements (Vorwerk et al., 1999), ossicle movement (Salih et al., 2016), the biomechanics of the incudomalleolar-joint (Ihrle et al., 2015) and the ossicles displacement (Hüttenbrink, 1988) (Lauxmann, 2012). The applied pressure was usually less than 0.5 kPa. The existing literature about middle ear pressure only relates medium static pressure changes to ossicular displacement and movement. To the best of our knowledge, the forces inside the OC resulting from static pressure have not been addressed. In the following temporal bone (TB) study we measure the forces in the OC provoked by quasi-static pressure at the TM. We use the term quasi-static because the applied pressure changes slowly with time but is constant for the duration of a single measurement. In our former study (Koch et al., 2016), which was about an implantable ISJ microphone sensor for a cochlea implant, we measured forces in the ISJ with a force measurement cell as a reaction force with the ossicular chain fixed at the point of measurement. For the current study we use the microphone sensor itself in a free floating design for measuring the force transferred directly through the OC without altering the OC behavior. Therefore, the current experiment closely represents realistic boundary conditions. Because of the measurement concepts’ novelty there was a lack of existing data for comparison and we decided to use 2 different types of sensors (piezo and strain gauge) for validation purposes in the quasi-static measurements. We examined the forces for a number of realistic scenarios for quasi-static pressure of up to 30 kPa as shown in Table 1. The study of quasi-static forces in the OC is interesting as a reference value for a variety of topics. In addition, the dynamic force in the ossicular chain provoked by sound pressure at the TM is also a physiological parameter with lots of practical implications. Measuring this force with the piezo sensor allowed for comparison to the METF (Middle ear transfer function as stapes velocity per ear canal sound pressure), which is the well-known core parameter for middle ear behavior (Zahnert, 2003). The accuracy of the dynamic measurement is restricted at higher frequencies due to the sensors own inertial mass. We furthermore applied a FE Model of the middle ear as introduced by (Bornitz et al., 2010) and (Oßmann, 2014) to support our analysis and verify the methods used.

was inserted as described by (Koch et al., 2016). After a posterior tympanotomy of the TB specimen, the ISJ was opened with a sickle knife, the long process of the incus was slightly lifted in direction of the longitudinal axis of the stapes with a needle and the sensor was slid between the long process of the incus and the stapes head with tweezers or a needle. Only TB specimen with intact stapedial tendons were chosen. A visual examination of the TBs was undertaken and only TB specimen with intact stapedial tendons were chosen. The sensor is held in place only due to friction forces, inertia, and the elastic force of the electric wire. Centered positioning of the sensor was controlled visually by microscope for the entire experiment. Quasi-static and dynamic forces inside the ISJ bend the sensor plate and therefore stretch the specific sensor element, which provokes a measureable charge change for the piezoelectric sensor and a change in the resistance for the SG element. The assembly and position of the sensor is schematized in Fig. 1 (A) and photographed in Fig. 1 (B).

2.2. Measurement setup 2.2.1. Sensor calibration The SG sensor was used for quasi-static measurements only; the piezoelectric sensor was used for dynamic measurements as well. The SG sensor was accompanied by a second dummy sensor of identical design for temperature compensation. Both were connected to a data acquisition board (NI PXI-4496, National Instruments Corporation, USA) equipped with a Wheatstone-halfbridge circuit for the SG measurement. The piezoelectric sensor signal was preamplified by a voltage amplifier (SR 560, Stanford Research Systems, USA) for the dynamic measurement and with a charge amplifier (Kistler Charge Amplifier 5018, Kistler Holding AG, Switzerland) for the quasi-static measurement. The calibration for the quasi-static measurements was undertaken by a force measurement cell (KA-S 0.5 A.S.T. Angewandte

2. Methods and material 2.1. Force sensor configuration and assembly The force sensor consists of an oval cylinder-shaped housing 4.5 mm long, 2.5 mm wide and 0.6 mm high. One lid of the housing is a 25 mm thin titanium plate which can be flexibly bent. Either a single crystal PMN-PT (Lead Magnesium Niobate-Lead Titanate) piezo or a semiconductor strain gauge (SG) element (strain gauge SS-060-033-500P and glue Omegabond 101, ME-Meßsysteme GmbH, Germany) are bonded beneath the thin plate. The sensor

Fig. 1. (A) Measurement principle and (B) Examplary Sensor in TB: Opening of the ISJ and insertion of a titanium sensor (3) between the incus (1) long process and stapes (2), equipped with a sensing element (single crystal piezo or strain gauge (SG) element) bonded beneath a thin plate which is part of the sensors housing. The contact of the bending plate to the long process of the incus allows to measure forces inside the ISJ. The external design of the sensor with piezo or SG is the same, the measurement with SG sensor is complemented with an additional identically sensor positioned close in the tympanic cavity for temperature compensation.

Please cite this article as: Koch, M et al., Static and dynamic forces in the incudostapedial joint gap, Hearing Research, https://doi.org/10.1016/ j.heares.2019.02.004

M. Koch et al. / Hearing Research xxx (xxxx) xxx

System Technik GmbH, Germany), coupled to the sensor with a needle. The needle was equipped with a circular head replicating the subsequent contact area between incus and the sensor plate's center as presented by (Chien et al., 2009). The sensor was temporarily fixed to a large metal block using instant adhesive. The force measurement cell was mounted on a piezoelectric step motor (Eppendorf Micromanipulator 5171, Eppendorf AG, Germany) through which a ramp force of up to 200 mN was applied over a time of approximately 30 s. The sensor signal was linearly proportional to the applied force if the step motors velocity was slow enough to avoid hysteresis effects. These were avoided by resetting the sensor signal every time the pressure dropped back to 0 Pa. This was possible because the drift effects don't change the linear relation of force onto the membrane versus sensor output for the piezo sensor respectively the resistance-change of the strain gauge sensor. Short-term pressure measurements were corrected for drift by linear interpolation. The calibration for the dynamic measurements was undertaken in a small acoustic chamber with cubical internal dimensions of 1 cm edge length. Acoustic pressure was applied by an earphone (Shure 315, Shure Inc., USA) as the excitation source. The stimulation level for the dynamic measurements of the calibration in the acoustic chamber was 94 dB SPL. A probe microphone (ER-7C, Etymotic Research Inc., USA) was used to supply a reference signal for the sound pressure. We used a multisine signal with sampling rate of 12 kHz and a block size of 1024 samples. The setup allowed for a constant pressure of around 1 Pa without disturbing effects such as standing waves in the range up to 5 kHz. The pressure of around 1 Pa was referenced by a probe microphone. The sensor was merely fixed with modelling clay on its backside inside the chamber. Only the sensor side with the sensing element beneath the membrane was thin enough to be influenced by the pressure. Note that in this setup, the force applied to the sensor was an area load on the whole membrane, as opposed to the incus contact which more closely resembles a point load. The FE-model of the sensor introduced in section 2.2.4 was used to convert the surface load to the desired point load: The FE model was excited with an area load and a point load. From the corresponding signal yields (in the acoustic frequency range), the scalar conversion factor to match the two loading conditions was derived. The factor was not frequency dependent. The SNR of the sensor in the dynamic measurements with excitation of 94 dB SPL at the tympanic membrane for a frequency range of 250 Hz to 6 kHz is 40 dB. The measuring threshold for both sensors (piezoelectric and strain gauge) for static measurements is around 2 mN. 2.2.2. Quasi-static force measurement For the quasi-static force measurement, the sensors were inserted into TBs as described in 2.1 after measurement of METF (see 2.2.3). The hole for the probe microphone was closed thereafter. The second SG sensor was placed in the tympanic cavity as close as possible to the ISJ for temperature compensation. The SG sensor was used in 5 TBs and the piezo sensor in 12 TBs. An overview of the TBs used is given in Appendix Table 4 for clarification. A step-motor-driven pressure pump was connected to the ear canal to apply quasi-static pressures. A manometer was connected as a reference. The initial measurement point at a pressure of 0 Pa constitutes the clamping-force, the force that acts on the sensor directly after insertion because the sensor has to push apart the long process of incus and stapes head. Pressures of up to 20 kPa were applied, limited by the air tightness of the tubes, the connection between the tubes and the ear canal, as well as the TM (which is already prone to rupture at much smaller pressures (Mirza, 2005). As it turned out to be

3

difficult to maintain the high pressures over a longer period, every measured data point was accessed separately always starting from 0 Pa pressure. This resulted in approx. 5 data points per measurement series. The piezo sensor has been introduced previously by (Koch, 2013) and (Koch et al., 2016). During the whole experiment, the position of the contact point between the long processes of the incus and the sensor was observed visually. Our conclusion from the former study was that a maximum position variation of 0.25 mm seems more reasonable if the surgeon is aware of the problem. This leads to variabilities of distinctly less than 5 dB. The current design is only smaller in height, which should not alter these results.

2.2.3. Dynamic force and METF measurement For the dynamic force measurements, a standard METF measurement as described in (Neudert et al., 2009) was performed for the frequency range of 250 Hz to 6 kHz. The reference pressure was measured by a probe microphone (ER-7C, Etymotic Research Inc., USA) placed 3 mm in front of the TM inserted through a small drilled hole in the anterior auditory canal wall. The sound excitation was realized with a sound generator (ER-2, Etymotic Research Inc., USA) in the ear canal. The stimulation was chosen identical to the calibration with a multisine signal with sampling rate of 12 kHz, a block size of 1024 and 94 dB SPL (but this time at the TM). The stapes footplate movement was measured at the center of the footplate with a laser Doppler vibrometer (LDV) (CLV 700, controller CLV 1000, Polytec GmbH, Germany). After the insertion of the piezoelectric sensor, the frequency-dependent relation between sensor signal-yield and dynamic sound pressure at the TM could be derived as explained in 2.1. With the conversion from the sensor calibration a dynamic force could be calculated. Dynamic forces were measured in 11 TBs. After each TB measurement (whether static or dynamic) the ISJ was closed and the METF measurement was repeated to confirm that the measurements were valid. Together with the quasi-static measurements, this study contains METF measurements of 26 TBs with intact ossicular chain. For the calculation of statistic magnitude ranges (see section 2.2.3) the METF measurement data was extended with an additional 18 frozen and defrosted TB METF measurements from our lab which were previously obtained with the same measurement setup and procedure as described above. This resulted in a total of 44 TB METFs for the statistical analysis.

2.2.4. Finite elements model Several steps in our analysis of the results below use a finite elements model of the sensor in the middle ear. The model of the middle ear (Bornitz et al., 2010) and (Oßmann, 2014) was complemented by a model of the force sensor built from simple 3D-20node solid elements for the housing and 3D-20-node-coupled field elements with electromechanical coupling for a single crystal piezo as shown in (Koch et al., 2016). The force measured by the sensor in the middle ear model is calculated to verify the dynamic force measurement. To examine whether the force is influenced by the sensor itself, the ISJ dynamic force in the model without sensor is also calculated. First it is examined as a ISJ node force to examine the anticipated force for a free moving OC, second it is examined as a reaction force at the fixed long process of the incus to achieve the force in a stiffened OC with fixed stapes. For validation purposes, we further compared results for a sensor housing that is completely fixated versus the sensor housing touching the stapes head area. The effect of the preload on the sensor and the load scenarios on the sensor have been examined through the comparison between linear and nonlinear simulations.

Please cite this article as: Koch, M et al., Static and dynamic forces in the incudostapedial joint gap, Hearing Research, https://doi.org/10.1016/ j.heares.2019.02.004

4

M. Koch et al. / Hearing Research xxx (xxxx) xxx

2.3. Further analysis of the results 2.3.1. Extrapolation of quasi-static forces The quasi-static forces could only be measured for pressures of 20 kPa or less. However, as mentioned in Table 1, relevant load scenarios range up to 30 kPa. The measurements show a linear relationship between measured force and applied pressure if both are plotted on a logarithmic scale. This can be explained by a stiffening of the anatomic structures for pressures at the TM up to 5 kPa. For further increase of pressure, the stiffening only increases marginally. We extrapolated all measurements reaching values of 10 kPa and beyond with linear fitting of the last 2 values in a double logarithmic scale up to 30 kPa. 2.3.2. Calculation of statistic magnitude ranges The distribution of the quasi-static and dynamic force measurements was logarithmically normally distributed and could be handled like a normal distribution after normalization. In the literature, the validity of the METF of a single TB measurement is usually compared to the METF range represented by (Rosowski et al., 2007) and (ASTM, 2005). This range was calculated as twice the standard deviation of the means from several studies. As stated by (Morse et al., 2018) this method is neither eligible for determining whether a single METF-measurement belongs to the 95% portion of the population nor for describing the whole population. Inductive statistics instead of descriptive statistics need to be applied to gain evidence for the population's properties. We propose the use of tolerance intervals (TI). TIs predict the limits of a given proportion from the whole population for a defined confidence interval (CI) and are calculated from a measured sample group. Statistical calculations were implemented using the programming-language “R” and its package ‘tolerance’ as described in (Young, 2010). The range of dynamic forces and METFs was determined by calculating the two sided 90% TI with p ¼ 0.05. For the quasi-static force measurements one-sided 95% TI limits with p ¼ 0.05 were calculated to define worst case scenarios. All mean measurements were validated by calculating the 95% CI of the mean (Weib and Bucsky, 2005). To assess the bandwidth of possible displacement, velocity and acceleration sensors in the OC from the METF measurements, the range width for a fixed frequency stays the same. The change in frequency-dependent progression can be calculated by converting the measured velocity to displacement or acceleration. 2.3.3. Dynamic range of different physical quantities Different physical quantities which can be measured at the ossicular chain show a different dynamic range. We demonstrate this for the kinetic quantities (displacement, velocity, acceleration) measured at the stapes footplate and the force measured in the incudo-stapedial joint. For the kinetic quantities just velocity is measured by the LDV and displacement and acceleration are computed thereof. All quantities are measured as transfer functions, related to the applied sound pressure at the TM. The dynamic range is the quotient between the maximum and minimum magnitude value of the physical quantity in a given frequency range, which we set to 250e6000 Hz. We use the mean of the measured data of each quantity to show the basic characteristics. In order to visualize and compare the dynamic range of the different quantities, they each have been normalized to the magnitude at 1 kHz. To obtain the whole dynamic range, the values obtained for the mean have to be superimposed by the interindividual variation described in terms of Tolerance intervals as explained in the previous section.

2.3.4. Assessment of the relationship between several results To evaluate the relationship between several measurement quantities some results were compared in boxplots. The relevance of differences is determined with a Welsh Two Sample T-Test. It is documented with 95% CIs for the mean difference and a p-value to conclude if the null hypothesis (no significant difference between sample groups) can be rejected. However as advocated by (Dahiru, 2008) and (Altman and Krzywinski, 2016) specific alpha-values like 0.05 to explicitly divide the results in significant or not significant are not defined to allow for small distinctions with weaker evidence. The evaluated data were single frequency points from 250 Hz to 4 kHz taken from the multisine measurements. The statistical analysis is limited to 4 kHz because the data at 6 kHz were at the boundary of the Nyquist-frequency and furthermore our analysis is focused on measurements of the stapes in piston-like motion only, which is increasingly invalid at higher frequencies. 3. Result and discussion 3.1. Static forces in the ISJ and extracted load scenarios The initial clamping force after sensor insertion, as measured in 16 TBs, was 23 mN þ - 18 mN. Fig. 2(A) and (B) show the relation of clamping force and METF. The clamping force measured for opening of the ISJ by about 0.6 mm was higher than in previous measurements with a force measurement cell and split TBs shown in (Koch et al., 2016, p. 20) where the clamping force was about 10 mN. Note that due to a slow drift in the measurement values, changes in clamping force that occurred at a similar or slower pace (over several minutes or more) could not be assessed. Within the examined timeframe, no relaxation or other changes in clamping force were apparent. We therefore evaluated only the forces measured at the time of insertion. The clamping forces were divided into two groups separated at F ¼ 12 mN. The boxplot (B) in Fig. 2 seems to indicate a correlation between larger clamping forces and the METF magnitude at higher frequencies (f > 1 kHz). The evidence for this is weak, with a t-test result for a 95% CI of 25 dB to 6 dB and a p-value of 0.1993 at 2 kHz and a 95% CI of 20 dB to 3 dB and a p-value of 0.124 at 4 kHz. Nonetheless, it seems reasonable to assume that a firmer or stiffer OC (higher force) could cause an upwards shift of the natural frequency of stapes motion, meaning that signal loss due to rocking motion or other complex modes may be diminished. Studies from (Gerig et al., 2015) on the stapes spatial motions with immobilized incudomalleolar joint support this hypothesis. Literature describes the ISJ as a synovial joint with a cartilage capsule filled with a synovial fluid (Zhang and Gan, 2011) and with an articular disk (Karmody et al., 2009). Because we opened the joint without removing parts like the disk or cartilage, the necessary stretch for insertion of the sensor is always its own thickness of 0.6 mm. The results of the pretension measurements are in accordance with the surgeons’ subjective impression of OC tightness. As described above, the maximum pressure applied in each experiment was limited by the airtightness of the specimen's ear canal. We were able to apply pressures of 10 kPa or more to 4 of 5 TBs for the SG measurement and to 8 of 11 TBs for the piezo sensor measurement (see Fig. 3). We observed large deformations of TM and especially incudomalleolar joint for the large static pressures. However, we observed no in-plane motion of the long process of the incus or the stapes head. The measured quasi-static forces were logarithmically normally distributed. To assess the relationship between quasi-static pressure at the TM and the force in the OC, the initial clamping forces were subtracted from the forces measured with TM quasi-static pressure. We observe a

Please cite this article as: Koch, M et al., Static and dynamic forces in the incudostapedial joint gap, Hearing Research, https://doi.org/10.1016/ j.heares.2019.02.004

M. Koch et al. / Hearing Research xxx (xxxx) xxx

5

Fig. 2. (A): Measured 17 TB METFs and related clamping forces. Grey area: METF Tolerance-Interval-range from 44 METF comparison measurements, see chapter 2.3.2 and Fig. 4. The METFs are classified regarding the corresponding clamping force after insertion. (B): Boxplot for assessment of the influence of sensor-clamping-force in TB's initial METF.

Fig. 3. (A) ISJ forces for quasi-static pressures at the TM (n ¼ 17). Dotted lines depict extrapolation up to 30 kPa for every TB which was measurable up to a minimum of 10 kPa. The Tolerance-Interval (grey) gives the prospective 95%-proportion of the population for worst-case scenarios. (B) Boxplot for comparison of quasi-static force measurement with SG sensor in frozen and defrosted TBs (n ¼ 5) vs. piezo sensor in fresh TBs (n ¼ 12).

logarithmic relationship between ISJ force and the quasi-static pressure. The mean force increases from 6 mN for 1 kPa pressure to 98 mN for 30 kPa pressure. The upper TI ranges from 54 mN for 1 kPa at load scenarios like small altitude variations to 926 mN for 30 kPa at the highest load scenario, 3 m diving. The measured forces for the different load scenarios are listed in Table 2. A possible application of these results is in the development and handling of ossicular prostheses. The influence of prosthesis length on AL stiffening is significant as stated by (Neudert et al., 2016). Fixed prostheses without flexible elements transfer atmospheric pressure changes directly to AL stiffening. Attempts are being made to develop flexible prostheses which can reduce the effect of timevarying circumstances (Stoppe et al., 2018). The results of the quasi-

static measurement could help to develop such prostheses by providing physical boundary conditions to be used in finite element models of the prostheses as well as physical models of the middle ear in addition to TB measurements. The results show that prostheses should be able to handle forces of around 50 mN in normal working mode, ideally reducing their effect on the AL whilst being robust enough to withstand maximum forces of up to 1 N. 3.2. Dynamic forces in the ISJ The FE-model calculation showed that 1 Pa surface load on the sensor membrane is equivalent to a point load of 3.968 mN. In the examined range, this is a linear, scalar, frequency-independent

Table 2 Measurement results: Forces for load scenarios of quasi-static pressures from Table 1. Load scenario

ISJ force mean

ISJ force max (95% TI one-sided)

Regular pressure variations like air pressure variations due to weather conditions or small altitude changes Valsalva's maneuver 2 m diving Aircraft landing 3 m diving

8 mN 37 mN 99 mN 118 mN 136 mN

54 mN 195 mN 637 mN 780 mN 926 mN

Please cite this article as: Koch, M et al., Static and dynamic forces in the incudostapedial joint gap, Hearing Research, https://doi.org/10.1016/ j.heares.2019.02.004

6

M. Koch et al. / Hearing Research xxx (xxxx) xxx

Fig. 4. (A) Dynamic force in the ISJ gap (n ¼ 11) with 2-sided 90% TI and comparative FE results of the middle ear with and without sensor. The width of the tolerance interval is depicted for several frequencies as well as the margin between maximum and minimum mean value to allow for comparability of the results variability. The modeled forces without sensor analyses solely the direction perpendicular to the stapes footplate. (B) Influence of TB condition “frozen and defrosted” vs.” fresh” in dynamic force measurement outcome. (C) Influence of static sensor-clamping-force in dynamic force measurement outcome.

factor. The mean dynamic force measured in the TB specimen is 26 mN between 500 Hz and 6 kHz (Fig. 4 (A)). Calculating means for every frequency in this range, The mean frequency response is quite flat with only 4.7 dB variation. The interindividual variation, calculated as 2-sided 90% TI, rises from 12 dB at low frequencies up to 25 dB at high frequencies. Sound excitation at 250 Hz was difficult to achieve using our setup, these values were therefore omitted from evaluation. In Fig. 5 the METF ranges of 44 TBs (26 from this study and 18 existing ones) are depicted. The interindividual range of 32 dB (low frequencies) to 38 dB (high frequencies) is much higher than for the force measurements. In addition, the mean METF, noted as displacement magnitude per pressure, drops by about 33 dB from 500 Hz to 6 kHz. This means that an OC displacement sensor at the stapes or stapes footplate must be able to handle input values with a variation of 64 dB for a fixed sound level. This has a great impact on sensor design, because it dramatically increases the sensor's

Fig. 5. Range of METFs of 44 TB specimen for comparison between the variability of METFs and dynamic force measurement. The mean (black) of the 44 METFs (coloured) and its confidence interval (green) are in accordance to the ASTM standard (dashed blue) but its range calculated as 2-sided 90% tolerance interval (grey) spans much wider.

target dynamic range. This variation could be slightly overestimated because the LDV measurements were not anglecorrected. The angle of LDV relative to the long axis of the stapes varied between 20 and 45 . The resulting error is 7%e29%, calculated by cosine correction, or 2.5 dB variance error, which is negligible. Fig. 6 shows the mean dynamic range of different physical quantities which can be measured at the ossicular chain. These are the displacement, velocity and acceleration at the stapes and the force transmitted in the incudo-stapedial joint. The magnitudes of each quantity are normalized to the value at 1 kHz and presented in a dB scale (see description in section 2.3.3). While the force exhibits a flat frequency-response characteristic, the displacement, the

Fig. 6. Normalized Magnitude in dB. Frequency-response characteristic of displacement (from 11 TBs from Fig. 4 (A)), velocity and acceleration at the stapes and the force (from 44 TBs from Fig. 5) transmitted in the incudo-stapedial joint (means in each case). ISJ force is based on dynamic force measurements in 11 TBs from Fig. 4 (A)); kinetic quantities of stapes are based on METF measurements in 44 TBs from Fig. 5. (Velocity was measured with LDV and displacement and acceleration were calculated thereof.) All quantities are normalized to their magnitude at 1 kHz. The dynamic range is the difference between maximum and minimum value of each graph.

Please cite this article as: Koch, M et al., Static and dynamic forces in the incudostapedial joint gap, Hearing Research, https://doi.org/10.1016/ j.heares.2019.02.004

M. Koch et al. / Hearing Research xxx (xxxx) xxx Table 3 Signal input range of means for several measurement principles of prospective sensors between f ¼ 250 Hz and f ¼ 6 kHz. Measurement principle

Dynamic range of means

Full dynamic range of means and interindividual variation calculated in terms of 95% TI range

Force Displacement Velocity Acceleration

5 dB 33 dB 16 dB 24 dB

40 dB 70 dB 53 dB 61 dB

velocity and the acceleration decrease at high and low frequencies, respectively. Correspondingly the dynamic range for the force is 5 dB (for the mean) and is much smaller than that for the stapes displacement (33 dB), velocity (16 dB) and acceleration (24 dB). To obtain the whole dynamic range the range of mean is superimposed with the interindividual range from the tolerance interval and listed in Table 3. The main finding is that the whole dynamic range of the force measurement is much smaller than that of the kinetic quantities. These findings are remarkable for the configuration of future AMEI sensor concepts. The relation between the dynamic force measurement and the initial clamping force is shown in Fig. 4, boxplot (C). Measurements have been categorized by clamping force F<10 mN or F>10 mN so as to obtain groups of roughly equal number of TBs. There is no discernible difference between groups at higher frequencies (f  1 kHz), and there is no clear evidence for a difference at lower frequencies (the T-test at 250 Hz shows a 95% CI of 19.8 dB to 2.6 dB with p ¼ 0.12). The results indicate that a higher pretension (stiffer OC) while diminishing the low frequency METF improving low frequency force transmission.

3.3. Validation As described above (section 2.3.2), there is cause for doubt as to whether the validation of TBs based on a comparison of their METFs to the (ASTM, 2005) standard is plausible. We have proposed a new method to describe the “normal” range of METFs, in a TB study but this method cannot be used for the 26 TBs in this study because the calculated tolerance intervals are themselves derived in part from the same TBs. Nevertheless a subjective comparison to the additional 18 TB METF measurements and to the (ASTM, 2005) standard, and also a visual and haptic inspection provided by the surgeons suggest that the TBs used were unscathed and can be considered representative of normal biomechanic functionality. We categorized the results of the quasi-static force measurement into measurements with the SG sensor in frozen and defrosted TBs versus piezo sensor in fresh TBs as shown in boxplot (B) of Fig. 3. TB 6 (measured with piezo sensor in frozen and defrosted TB) had to be excluded for this analysis. The t-test difference between the quasi-static measurements is distinct for the clamping force only with 95% CI ¼ 2.8 dBe15.3 dB and p ¼ 0.01, negligible for a quasi-static pressure of 1 kPa and 5 kPa with pvalues of 0.36 and slightly remarkable for a quasi-static pressure of 10 kPa with 95% CI ¼ 0.7 dBe14.4 dB and p ¼ 0.03. On first sight the analysis indicates a small change only for the clamping force (p ¼ 0 kPa). Based on the available data, no clear difference could be discerned between the two sensor types, this confirms the targeted validation of the piezo sensor measurements with help of the strain gauge measurements. There was also no relevant difference in measured force between the TBs status fresh or frozen and defrosted for quasi-static pressure excitation of p > 0 kPa. In Fig. 4, boxplot (B), the dynamic force measurement results for frozen and

7

defrosted TBs versus fresh TBs are compared. Even more than for the quasi-static measurements, we see no relevant difference. This indicates that dynamic force measurements can be undertaken for frozen and defrosted TBs as well and there is no need to take fresh TBs. The same finding was proposed by (Ravicz et al., 2000) for the LDV measurement of METF. We validated the sensor calibration concept and the measurements with the FE model. For static single load force excitation FE simulations showed no difference if the stapes side of the housing is fixed on a small centered circular area like in middle ears or the whole sensors backside like in the calibration. This is to be expected because the housing is much more stable than the thin sensing membrane. FE comparisons between linear and nonlinear simulation show that the sensor response was not dependent on the preload, respectively the load scenarios for the examined frequency range and applied loads up to 30 kPa at the TM. The modelling of the dynamic forces in the ISJ as shown in Fig. 4A) is in good accordance to the measurement data for up to 3 kHz. Beyond that frequency the FE model is not validated (Bornitz et al., 2010) and therefore not depicted. In the FE model the sensor insertion does not substantially alter the results. 3.4. Transferability of the results While the boundary conditions are slightly altered due to the widening of the ISJ gap in our setup, larger modifications, like splitting the TB or removal of the cochlea, can be avoided. An even thinner sensor would be desirable to diminish the effects of widening. However, for the quasi-static measurements, subjective observations of ossicle and TM movement suggest that the sensor insertion only partly affects the pressure protection gliding-joint effect of the incudomalleolar joint pointed out by (Hüttenbrink, 1988). Sensor insertion not only affects the incudomalleolar joint, it also stiffens the AL. FE-simulations of a free OC versus a completely fixed long process of the incus show a variation of the dynamic forces up to about 6 dB (see Fig. 4 (A)). (Lauxmann, 2012) suggests the compensation of amplitude and phase differences between the manometer and the TM as the point of application for the quasi-static pressure. This was not realized in this study. We postulate this correction is not necessary for our measurement setup with its slow pressure application. The measurement position of the sensor can be adjusted very accurately, because differences in the positioning perpendicular to the long stapes axis can be easily seen and corrected. The maximum anticipated deviation is 0.25 mm from the center, see chapter 2.2.2. Because there is no further joint between ISJ and AL the measured force should be applicable with good precision to the AL for the quasi-static measurements. Former studies by (Fisch and May 1994) suggest that the opening and subsequent restoration of the ISJ does not alter the function of the OC, while (Farahmand et al., 2016) states that a high frequency hearing loss might occur. However, while the ISJ in its original state is also capable of transferring lateral movements, the joint separation and insertion of a planar surface transforms the joint into something like a rotational joint. Therefore, the degree of motion is extended. For quasi-static measurements and dynamic measurements inside a frequency range with a dominant stapes piston motion (up to 1 kHz (Hato et al., 2003),) this should be negligible. Beyond these frequencies the validity of the findings could be limited. The synovial fluids principle of operation inside the joint examined by (Jiang and Gan, 2018) is impeded by the opening of the joint. Therefore, the proposed frequency dependency of ISJ stiffness could be diminished in our experiment. Several mechanisms of TM, ossicles, mastoid and Eustachian tube help to protect the middle and inner ear by regulating and

Please cite this article as: Koch, M et al., Static and dynamic forces in the incudostapedial joint gap, Hearing Research, https://doi.org/10.1016/ j.heares.2019.02.004

8

M. Koch et al. / Hearing Research xxx (xxxx) xxx

buffering incoming quasi-static pressures as stated by (Alper et al., 2011; Gaihede and Kabel, 2000; Padurariu et al., 2015). Possibly not all of these in-vivo mechanisms can be reproduced adequately in TB measurements. Most of the discussed restrictions tend to result in a slight overestimation of the measured forces. Therefore, the derived quasi-static and dynamic forces can be interpreted as worst-case scenarios. 4. Conclusion In this study, quasi-static forces and dynamic forces in the ISJ were examined. The results are applicable for assessing the load onto the AL as well as AMEI sensors or prostheses in the OC. In FE simulations and experiments on prostheses, the values found could be used as input parameters which closely resemble actual conditions. The dynamic measurements show that the measurement of force in the OC exhibits much smaller variations than the METF. 5. Author contribution Study design, involvement in sensor design, definition of quasistatic load scenarios, measurement execution, FE simulations and data evaluation and discussion were by Martin Koch. Coding of measurement software for dynamic force measurement and execution of dynamic force measurement were by Till Moritz Ebinger. Assembly of sensors and calibration of dynamic force measurement, measurement execution of dynamic force measurement, definition of quasi-static load scenarios were by Martin Angerer. Coding of measurement software for quasi-static force measurement and calibration of quasi-static force measurement were by Thomas Stoppe. Results discussion, assessment of transferability of METF-range in displacement/velocity/acceleration and participation in technical design of dynamic force sensor were by Matthias Bornitz. Execution of the surgical part of the TB experiments and assessment of the results’ medical relevance were by Marcus Neudert. Evaluation of study methodology from a medical/ surgical point of view was by Thomas Zahnert. Acknowledgements The study was part of a third-party project funded by MED-EL Medical Electronics, Austria. In addition, the authors want to thank Hannes Seidler, Steffen Oßmann, Nicoloz Lasurashvili, Marie-Louise Metasch, Sandra Glausch and Christoph Müller for their valuable contributions. (all ERCD Ear Research Center Dresden, Germany) Appendix

Table 4 Temporal bones used Experiment

Sample size

TB Nr.

TB condition

Clamping force and static force in relation to TM-pressure SG sensor 5 TB1 to TB5 Frozen and defrosted Piezo sensor 12 TB6 to TB6 frozen and defrosted, TB7 to TB17 TB16 fresh Dynamic force in relation to TM-pressure Piezo sensor 11 TB16 to Fresh TB22 TB23 to Frozen and defrosted TB26 METF comparison 18 TB27 to Frozen and defrosted data TB41

References Alper, C.M., Kitsko, D.J., Swarts, J.D., Martin, B., Yuksel, S., Cullen, D., Villardo, R.J.M., Doyle, W.J., 2011. Role of the mastoid in middle ear pressure regulation. Laryngoscope 121, 404e408. https://doi.org/10.1002/lary.21275. Altman, N., Krzywinski, M., 2016. Points of significance: P values and the search for significance [WWW Document]. Nat. Methods. https://doi.org/10.1038/nmeth. 4120. ASTM, 2005. Standard Practice for Describing System Output of Implantable Middle Ear Hearing Devices. Bornitz, M., Hardtke, H.-J., Zahnert, T., 2010. Evaluation of implantable actuators by means of a middle ear simulation model. Hear. Res. 263, 145e151. https://doi. org/10.1016/j.heares.2010.02.007. Chien, W., Northrop, C., Levine, S., Pilch, B., Peake, W., Rosowski, J., Merchant, S., 2009. Anatomy of the distal incus in humans. J. Assoc. Res. Otolaryngol. 10, 485e496. https://doi.org/10.1007/s10162-009-0179-6. Dahiru, T., 2008. P e Value, a true test of statistical significance? a cautionary note. Ann. Ib. Postgrad. Med. 6, 21e26. €o €sli, C., Ulku, C.H., Farahmand, R.B., Merchant, G.R., Lookabaugh, S.A., Ro McKenna, M.J., de Venecia, R.K., Halpin, C.F., Rosowski, J.J., Nakajima, H.H., 2016. The audiometric and mechanical effects of partial ossicular discontinuity. Ear Hear. 37. Feldmann, H., 1973. Physiologie und Pathophysiologie der Mittelohrventilation. Teil 2. Z Laryng Rhinol 52, 555e572. Fisch, U., May, J., 1994. Tympanoplasty, Mastoidectomy and Stapes Surgery. Georg Thieme, Stuttgart, New York. Gaihede, M., Kabel, J., 2000. The normal pressure-volume relationship of the middle ear system and its biological variation. In: Rosowski, J.J., Merchant, S.N. (Eds.), Second International Symposium on Middle-Ear Mechanics in Research and Otosurgery. Kugler Publications, The Hague, The Netherlands, Boston, MA, USA, pp. 59e70, 1999. €o €sli, C., Dalbert, A., Dobrev, I., Pfiffner, F., Eiber, A., Huber, A.M., Gerig, R., Ihrle, S., Ro Sim, J.H., 2015. Contribution of the incudo-malleolar joint to middle-ear sound transmission. Hear. Res. 327, 218e226. https://doi.org/10.1016/j.heares.2015.07. 011. Hato, N., Stenfelt, S., Goode, R.L., 2003. Three-dimensional stapes footplate motion in human temporal bones. Audiol. Neuro. Otol. 8, 140e152. https://doi.org/10. 1159/000069475. Hüttenbrink, K.B., 1988. The mechanics of the middle ear at static air pressure. Acta Otolaryngol Suppl 451, 1e36. €o €sli, C., Sim, J.H., Huber, A.M., Eiber, A., 2015. Ihrle, S., Gerig, R., Dobrev, I., Ro Biomechanics of the incudo-malleolar-joint e experimental investigations for quasi-static loads. Hear. Res. 340, 69e78. https://doi.org/10.1016/j.heares.2015. 10.015. Jiang, S., Gan, R.Z., 2018. Dynamic properties of human incudostapedial jointExperimental measurement and finite element modeling. Med. Eng. Phys. 54, 14e21. https://doi.org/10.1016/j.medengphy.2018.02.006. Karmody, C., Northrop, C., Levine, S., 2009. The incudostapedial articulation: new concepts. Otol. Neurotol. 30, 990e997. https://doi.org/10.1097/MAO. 0b013e3181b0fff7. €re Einführung in die Physik der Klose, B., 2008. Meteorologie: Eine interdisziplina Atmosph€ are. Springer-Lehrbuch. Springer-Verlag, Berlin Heidelberg. Koch, M., Essinger, T.M., Stoppe, T., Lasurashvili, N., Bornitz, M., Zahnert, T., 2016. Fully implantable hearing aid in the incudostapedial joint gap. Hear. Res. 340, 169e178. https://doi.org/10.1016/j.heares.2016.03.015. Koch, M., Seidler, H., Hellmuth, A., Bornitz, M., Lasurashvili, N., Zahnert, T., 2013. Influence of the middle ear anatomy on the performance of a membrane sensor in the incudostapedial joint gap. Hear. Res. 301, 35e43. Lauxmann, M., 2012. Nichtlineare Modellierung des Mittelohrs und seiner angrenzenden Strukturen. Institut für Technische und Numerische Mechanik Universit€ at Stuttgart. Lehnhardt, E., 2009. Praxis der Audiometrie. Georg Thieme Verlag. Looga, R., 2005. The Valsalva manoeuvre–cardiovascular effects and performance technique: a critical review. Respir. Physiol. Neurobiol. 147, 39e49. https:// doi.org/10.1016/j.resp.2005.01.003. Mirza, R.H., 2005. Otic barotrauma from air travel. J. Laryngol. Otol. 119, 366e370. https://doi.org/10.1258/0022215053945723. Morse, R., Mitchell-Innes, A., Prokopiou, A., Irving, R., Begg, P., 2018. Inappropriate use of the “Rosowski Criteria” and “Modified Rosowski Criteria” for assessing the normal function of human temporal bones. Audiol. Neurotol. (in press). Neudert, M., Bornitz, M., Lasurashvili, N., Schmidt, U., Beleites, T., Zahnert, T., 2016. Impact of prosthesis length on tympanic membrane's and annular ligament's stiffness and the resulting middle ear sound transmission. Otol. Neurotol. 37, e369ee376. https://doi.org/10.1097/MAO.0000000000001064. Neudert, M., Zahnert, T., Lasurashvili, N., Bornitz, M., Lavcheva, Z., Offergeld, C., 2009. Partial ossicular reconstruction: comparison of three different prostheses in clinical and experimental studies. Otol. Neurotol. 30, 332e338. https://doi. org/10.1097/MAO.0b013e31819679dd. Oßmann, S., 2014. Erstellung eines patientenspezifischen FE-Modells des Mittelohrs und dessen Ordnungsreduktion (Diplomarbeit). Technische Universit€ at Dres€t Maschinenwesen. Institut für Festko €rpermechanik. den, Fakulta Padurariu, S., de Greef, D., Jacobsen, H., Kamavuako, E.N., Dirckx, J.J., Gaihede, M., 2015. Pressure buffering by the tympanic membrane. In vivo measurements of middle ear pressure fluctuations during elevator motion. Hear. Res. 340,

Please cite this article as: Koch, M et al., Static and dynamic forces in the incudostapedial joint gap, Hearing Research, https://doi.org/10.1016/ j.heares.2019.02.004

M. Koch et al. / Hearing Research xxx (xxxx) xxx 113e120. Ravicz, M.E., Merchant, S.N., Rosowski, J.J., 2000. Effect of freezing and thawing on stapes-cochlear input impedance in human temporal bones. Hear. Res. 150, 215e224. Rosowski, J.J., Chien, W., Ravicz, M.E., Merchant, S.N., 2007. Testing a method for quantifying the output of implantable middle ear hearing devices. Audiol. Neuro. Otol. 12, 265e276. https://doi.org/10.1159/000101474. Salih, W.H.M., Soons, J.A.M., Dirckx, J.J.J., 2016. 3D displacement of the middle ear ossicles in the quasi-static pressure regime using new X-ray stereoscopy technique. Hear. Res., MEMRO 2015 e Basic Science meets Clinical Otology 340, 60e68. https://doi.org/10.1016/j.heares.2015.12.003. Samuels, M.P., 2004. The effects of flight and altitude. Arch. Dis. Child. 89, 448e455. https://doi.org/10.1136/adc.2003.031708. Stoppe, T., Bornitz, M., Lasurashvili, N., Sauer, K., Zahnert, T., Zaoui, K., Beleites, T., 2018. Function, applicability, and properties of a novel flexible total ossicular

9

replacement prosthesis with a silicone coated ball and socket joint. Otol. Neurotol. 39, 739. https://doi.org/10.1097/MAO.0000000000001797. Vorwerk, U., Steinicke, G., Begall, K., 1999. Observation of the eardrum movements during quasi-static pressure changes by high-speed digital imaging. Audiol. Neuro. Otol. 4, 150e155. https://doi.org/10.1159/000013834. Weiß, C., Bucsky, P., 2005. Basiswissen Medizinische Statistik, third ed. SpringerLehrbuch. Springer-Verlag, Berlin Heidelberg. Young, D.S., 2010. Tolerance: an R package for estimating tolerance intervals. J. Stat. Software. https://doi.org/10.18637/jss.v036.i05. Zahnert, T., 2003. [Lasers in ear research]. Laryngo-Rhino-Otol. 82 (Suppl. 1), S157eS180. https://doi.org/10.1055/s-2003-38931. Zhang, X., Gan, R.Z., 2011. Experimental measurement and modeling analysis on mechanical properties of incudostapedial joint. Biomechanics Model. Mechanobiol. 10, 713e726. https://doi.org/10.1007/s10237-010-0268-9.

Please cite this article as: Koch, M et al., Static and dynamic forces in the incudostapedial joint gap, Hearing Research, https://doi.org/10.1016/ j.heares.2019.02.004