Get 20M+ Full-Text Papers For Less Than $1.50/day. Start a 14-Day Trial for You or Your Team.

Learn More →

Experimental validation of a three-dimensional heat transfer model within the scala tympani with application to magnetic cochlear implant surgery

Experimental validation of a three-dimensional heat transfer model within the scala tympani with... Experimental validation of a three-dimensional heat transfer model within the scala tympani with application to magnetic cochlear implant surgery Fateme Esmailie, Mathieu Francoeur, Tim Ameel Abstract—Magnetic guidance of cochlear implants is a promis- a magnet in the cochlea may cause medical complications if ing technique to reduce the risk of physical trauma during the patient is exposed to a strong external magnetic such as surgery. In this approach, a magnet attached to the tip of the in magnetic resonance imaging. The removal can be accom- implant electrode array is guided within the scala tympani using a plished by melting the substance bonding the implant electrode magnetic field. After surgery, the magnet must be detached from array and the magnet via localized heating. Consequently, the implant electrode array via localized heating and removed from the scala tympani which may cause thermal trauma. the magnet detachment process may cause localized hyper- Objectives: The objective of this work is to experimentally thermia within the scala tympani, which is one of the more validate a three-dimensional (3D) heat transfer model of the serious impediments preventing the establishment of magnetic scala tympani which will enable accurate predictions of the cochlear implant surgery [4], [5]. Localized hyperthermia in maximum safe input power to avoid localized hyperthermia when the ear, which is not unique to magnetic cochlear implant detaching the magnet from the implant electrode array. Methods: Experiments are designed using a rigorous scale analysis and surgery, arises in a variety of applications such as infrared performed by measuring transient temperatures in a 3D-printed neural stimulation implants [6] – [10], stapedectomy [11], [12], scala tympani phantom subjected to a sudden change in its caloric test [13], [14], and radio-frequency radiation devices thermal environment and localized heating via a small heat such as cellular phones [15] – [17]. In addition, studies have source. Results: The measured and predicted temperatures are shown that localized hypothermia can preserve tissues after in good agreement with an error less than 6%. Conclusions: The validated 3D heat transfer model of the scala tympani is cochlear implantation [18], [19]. finally applied to evaluate the maximum safe input power to Despite the importance of localized hyperthermia and hy- avoid localized hyperthermia when detaching the magnet. For the pothermia in hearing research, validated three-dimensional most conservative case where all boundaries except the insertion (3D) heat transfer computational models within cochlear opening are adiabatic, the power required to release the magnet canals are scarce. Thompson et al. [6] – [8] and Liljemalm [20] attached to the implant electrode array by 1 mm of paraffin is approximately half of the predicted maximum safe input power. developed a heat transfer model to evaluate the thermal impact Significance: This work will enable the design of a thermally safe of infrared neural stimulation implants. In this technique, the magnetic cochlear implant surgery procedure. implant focuses infrared radiation for stimulating modulus Index Terms—Cochlear implant, Magnetic guidance, Scala nerves, which are a bundle of nerves located in the middle of tympani, Heat transfer simulations, Heat transfer experiments the bony cochlea. They validated their numerical results using the experimental data of Shapiro et al. [9], who conducted in-vitro experiments by stimulating an oocyte. Thompson et al. [6] – [8] and Liljemalm [20] analyzed the feasibility of infrared neural stimulation implants, but the actual geometry I. I NTRODUCTION of the cochlea was not considered in the validation phase of Magnetic guidance of cochlear implants is a promising these works. Similar to magnetic cochlear implant surgery, surgical technique that is expected to mitigate physical trauma infrared neural stimulation implant surgery involves localized arising with manual insertion of implants [1]– [3]. In this heating. However, neither the heat source nor the targeted technique, a magnetic field guides the implant electrode array tissues are similar to the magnet detachment process. The within the scala tympani, one of the cochlear canals, using a impact of therapeutic hypothermia in the cochlea was studied small magnet attached to the implant tip. After surgery, the by Tamames et al. [18] using a COMSOL Multiphysics model magnet must be detached from the implant electrode array validated with experimental data. The focus of their work was and removed from the scala tympani, since the presence of on the preservation of cochlear tissues via hypothermia and the design of a cooling device. Whereas this work is an excellent Submission date: 09/01/2020 Research reported in this publication was supported by the National Institute example of a validated heat transfer model within the cochlea, on Deafness and Other Communication Disorders of the National Institutes the application and physics are different from the magnetic of Health under Award Number R01DC013168. cochlear implant surgery problem. F. Esmailie is with Department of Mechanical Engineering, University of Utah, Salt Lake City, Utah 84112, USA. The objective of this paper is, therefore, to validate a M. Francoeur is with Department of Mechanical Engineering, University 3D heat transfer computational model of the scala tympani, of Utah, Salt Lake City, Utah 84112, USA . which will enable quantification of the risks of localized T. Ameel is with Department of Mechanical Engineering, University of Utah, Salt Lake City, Utah 84112, USA (e-mail: ameel@mech.utah.edu). hyperthermia during magnetic cochlear implant surgery. In arXiv:2009.07199v1 [physics.app-ph] 15 Sep 2020 2 Fig. 1. Cutaway view of the cochlea. The cochlea includes three canals, Fig. 2. 3D drawing of the scala tympani phantom. namely the scala tympani, the scala vestibuli, and the scala media. The cochlear implant electrode array is magnetically-guided in the scala tympani. physical scala tympani model used in the experiments will be hearing research, new surgical methods or treatments are tested referred to as the phantom. The phantom, consisting of a hol- in phantoms [21] – [23], cadavers [23], [24], or animals low scala tympani canal, was 3D-printed by Realize Inc. with [25] – [27]. Here, transient temperatures measured in a 3D- a transparent material called Somos WaterShed XC 11122. The phantom includes twelve ducts enabling the insertion of printed scala tympani phantom subjected to a sudden change thermocouples for measuring transient temperatures along the in its thermal environment and localized heating are used for scala tympani. A separate insertion opening included in the validating the computational model. As the numerical and phantom facilitates implantation of an implant (see Fig. 2). experimental results are in good agreement, the model is Transient temperatures are measured for two distinct cases, applied to determine the maximum safe input power to avoid namely when the phantom is subjected to a sudden change in localized hyperthermia when detaching the magnet. its thermal environment, and when there is localized heating The balance of the paper is organized as follows. Section within the phantom. The experimental data are then compared II is devoted to a description of the problem. Subsequently, the design of the experiment used to validate the heat transfer against numerical predictions obtained with the computational model and the associated scale analysis are discussed. Tran- heat transfer model having a geometry identical to that of the sient temperatures measured in a 3D-printed scala tympani phantom. phantom subjected to a sudden change in its thermal environ- A deterrent to thermocouple temperature measurement at ment and localized heating are compared against numerical small scales is the possible influence they may have on the predictions in section IV. Finally, the validated heat transfer measurement. If the size of the thermocouples is of the same model is used to predict the maximum input power to avoid order of magnitude as the phantom, the thermocouples may hyperthermia in the scala tympani for various boundary con- affect the temperature distribution and fluid flow within the ditions and heating durations. scala tympani canal, and may act as heat sinks. To avoid these potential issues, a scaled-up phantom is used in the II. DESCRIPTION OF THE PROBLEM experiments. This is discussed next. The inner ear consists of the cochlea and the organ of III. EXPERIM ENTAL DESIGN AND SCALE ANALYSIS balance [28]. The cochlea is a spiral bony organ with two and a half turns consisting of three canals, namely the scala vestibuli, The impact of phantom scale on the physics of the problem the scala media and the scala tympani (see Fig. 1) [29] – and the significance of each term in the energy and momentum [31]. The scala tympani and scala vestibuli are filled with a balance equations describing fluid and heat transport is studied dilute saline fluid called perilymph, while the scala media is via a scale analysis. The scale analysis is a well-established filled with endolymph which is also a dilute saline fluid [29] method in thermal-fluid science that enables eliminating negli- – [31]. Both the perilymph and endolymph are characterized gible physical phenomena in a process, and permits designing by thermophysical properties identical to that of water. The physically meaningful experiments. cochlear implant is magnetically guided in the scala tympani. The scale analysis requires non-dimensionalizing the energy and momentum balance equations. A general form of the Physical models of the scala tympani have been previously energy balance equation is given by [34]: constructed [32], [33]. In this work, the physical model devel- oped by Lisandro et al. [33] (see Fig. 2) is used for validating @T c + c urT +r (krT ) +r q + p p the computational heat transfer model within the scala tym- @t (1) pani. To avoid confusion with the computational model, the c ! (T T ) = q + q bl p;bl bl bl met 3 where q is the radiative heat flux vector, and q and q are heat transfer (diffusion). The energy balance in the perilymph met respectively the volumetric heat sources due to metabolic heat does not include the input power density, but accounts for heat generation and the magnet. Blood perfusion (last term on the transfer by conduction and convection. The energy balance left-hand side of (1) in the cochlea absorbs heat and contributes equation in the perilymph region is: to energy dissipation. The blood perfusion rate in the cochlea is @T ml less than 1 [8]. In infrared neural stimulation, the energy c + c urT +r (krT ) = 0 (3) ming p p @t dissipated by perfusion in the cochlea is negligible compared to the magnitude of the heat source [7], [9], [18], [19], [8]. where the second term on the left-hand side is the heat trans- For example, Thompson et al. [7], [8] discussed all possible port by convection. To calculate this term, the velocity vector biological cooling methods in the cochlea (perfusion, cochlear fluid flow, evaporation of water from tissues), and concluded should be simultaneously computed by solving the momentum balance (i.e., Navier-Stokes) equation in the perilymph: that these mechanisms are much lower than the magnitude of the laser heat source power (6.25 mW) used for infrared neural @u stimulation implants. Here, an input power of 21 mW applied +  (ur) u = r pI + g (4) @t for 10 s is necessary to detach the magnet by assuming that the magnet is attached to the implant electrode array by 1 where the first term on the left-hand side is the acceleration mm of paraffin having a melting point of 43 C. This is more term and the second term represents advection. The first term than three times larger than the power reported by Thompson on the right-hand side is the pressure force and the second et al. [7], [8]. The conclusions of Thompson et al. [7], [8] term represents the body force (gravity) acting in the negative may be applicable for magnetic cochlear implant surgery, but z direction. Note that the external magnetic field guiding heat dissipation by blood perfusion should also be compared the magnet through the scala tympani is turned off after the to other heat dissipation mechanisms in the scala tympani, and surgery, such that magnetohydrodynamic effects do not come not solely to the external input power. into the picture. According to [7], the heat removal rate per unit volume The momentum and energy balance equations are non- by perfusion in the ear for a temperature difference of 3 C ml dimensionalized hereafter for the scale analysis. The dimen- and a perfusion rate of 1 is  0.03 mW. For magnetic ming sionless Cartesian coordinates are defined as: cochlear implant surgery, the temperature difference may be up to 6 C, such that the perfusion heat removal rate is estimated x y z at 0.06 mW (assuming a cochlear volume of 185.4 mm X = ; Y = ; Z = (5) kW L L L c c c [35] and a heat removal rate of 0.33 ). In contrast, heat dissipation through the cochlear temporal bone characterized where L is a characteristic length. Similarly, the Cartesian by a surface area of 250 mm [36] and a thickness of 4 mm components of the velocity vector are non-dimensionalized as [37] is approximately 120 mW (assuming a cochlear volume follows: 3 kW of 185.4 mm [35] and a heat removal rate of 647.2 as u u u estimated by Fouriers law). As a result, the blood perfusion x y z U = ; U = ; U = ; (6) x y z term in (1) is negligible with respect to heat dissipation by u u u 0 0 0 conduction through the temporal bone. p where u (= g L (T T )) is derived by assuming that 0 c max i The metabolic heat generation rate per unit volume (q ) in met kW the gravitational and viscous forces are in equilibrium. The a cochlea is approximately 1.1 [18]. Under the assumption TT dimensionless temperature is defined as  = , where that the entire cochlea is filled with blood, which is an T T max i T is initial temperature, and the dimensionless time, also overestimation, the metabolic heat generation is  0.2 mW. called the Fourier number, is given by Fo = . This is two orders of magnitude smaller than the magnitude L Substituting the dimensionless parameters into (2) to (4), of the heat source (21 mW), such that the metabolic heat the dimensionless energy balance in the solid magnet region generation term can be neglected in (1). is: Heat transport by radiation is estimated at 0.01 mW using 2 2 2 @ @  @  @ the Stefan-Boltzmann law, which is three orders of magnitude = + + + Q (7) 2 2 2 @Fo @X @Y @Z smaller than the heat source. The contribution to heat transfer by radiation is therefore neglected in (1). qL After neglecting blood perfusion, metabolic heat generation, where Q = is the dimensionless input power. The V k(T T ) max i and radiation heat transfer, the simplified energy balance dimensionless energy balance in the perilymph is written as: equation applied to the solid magnet region where heat is @ @ @ @ 0:5 transferred solely by conduction is given by: + PrGr U + U + U = x y z @Fo @X @Y @Z (8) @T 2 2 2 @  @  @ c +r (krT ) = q (2) + + @t 2 2 2 @X @Y @Z g (T T )L s 1 The first term on the left-hand side of (2) represents thermal where Pr = and Gr = are the Prandtl energy storage, whereas the second term is the conduction and Grashoff numbers, respectively. Finally, the dimensionless 4 momentum balance equations in the perilymph along each that natural convection heat transfer (second term on the left- Cartesian coordinate are given by: hand side of (8)) is negligible in comparison to conduction (right-hand side of (8)). This result is in agreement with our 1 @U @U @U @U x x x x 0:5 previous study performed on an uncoiled cochlea model [4], + Gr U + U + U = x y z Pr @Fo @X @Y @Z where it was shown that natural convection has a negligible (9) 2 2 2 @ U @ U @ U x x x impact on the thermal management of the scala tympani during + + 2 2 2 @X @Y @Z magnetic cochlear implant surgery. As such, heat transfer by natural convection is neglected in the remainder of this paper. Note that the accuracy of this assumption is later confirmed 1 @U @U @U @U y y y y 0:5 + Gr U + U + U = x y z with experimental data. Pr @Fo @X @Y @Z (10) In addition to the two-concentric cylinder problem, the 2 2 2 @ U @ U @ U y y y accuracy of the scale analysis for a prototype and model (3:1) + + 2 2 2 @X @Y @Z of the scala tympani is also analyzed. The scale analysis is performed based on the geometry and thermophysical proper- 1 @U @U @U @U ties of the scala tympani phantom discussed in section II. A z z z z 0:5 + Gr U + U + U = x y z spiral-shaped heat source is implanted in the narrowest region Pr @Fo @X @Y @Z (11) 2 2 2 of the phantom, while it is assumed that all boundaries of the @ U @ U @ U z z z 0:5 + + + Gr phantom are adiabatic and that the initial temperature is 0 C. 2 2 2 @X @Y @Z The thermophysical properties of perilymph (same as water), Based on (7) to (11), the scale analysis implies that if the the phantom prototype material (Somos WaterShed XC11122), dimensionless parameters Fo; Pr; Gr; and Q are equal for the and the heat source (Ni-Cr80 wire) are listed in Table I. actual physical model (prototype) and the scaled-up physical The geometry of the scala tympani prototype is illustrated model (simply called model hereafter), then the dimensionless in Fig. 4 (a). For equivalent values of Q, the power inputs temperature  and dimensionless velocities U ; U ; U will be for the phantom prototype and model are 0.9 W and 0.1 W, x y z equal for the prototype and the model. respectively. Temperatures are calculated at a reference point The scale analysis is first verified by solving a simple (prototype: x = 0 mm, y = 0 mm, z = -2/3 mm, model: x = heat transfer problem between two concentric cylinders. The 0 mm, y = 0 mm, z = -2 mm) and a cut line on the cut plane inner cylinder is heated via constant heat rates of 0.01 W crossing the reference point (see Figs. 4, (b), (c) and (d)). (model) and 0.04 W (prototype). The boundary of the outer The transient dimensionless temperature at the reference point cylinder is adiabatic, and the gap region between the two is plotted in Fig. 5. As expected, the temperature increases cylinders is filled with water. The thermophysical properties with increasing time when the heat source is turned on. The required for the simulations are listed in Table I. The initial temperature gradient increases after some time and, as a result, temperature of the entire system is 20 C. For the prototype, the temperature increase rate is enhanced. This trend is the the diameter and length of the inner cylinder are 1 mm and 5 same for both the phantom prototype and model with less than mm, respectively. The outer cylinder has a diameter of 2 mm 5% difference between individual data. The differences are due and a length of 5 mm. Here, the model is two times larger to the disparate meshes for the prototype and model as well as than the prototype (see Fig. 3). The characteristic length is truncation and round off errors. The dimensionless temperature defined as the gap distance between the two cylinders. The along the cut line in the phantom prototype is plotted in velocity at the surface of the inner and outer cylinders is zero Fig. 6. The temperature is maximum near the heat source. because of the no-slip boundary condition. Consequently, the As the distance with respect to the heat source increases, dimensionless velocity is zero, and reaches a maximum value the temperature gradient decreases and, consequently, the heat between the two cylinders. The dimensionless velocities for transfer rate drops. The spatial temperature distribution trend the prototype and model are in a good agreement with less is identical for the phantom prototype and model with less than 1% difference. The temperature at the surface of the than 5% discrepancy at a few locations. Again, the differences heated inner cylinder is a maximum and decreases towards the are due to different mesh as well as truncation and round off outer cylinder. Again, differences between the dimensionless errors. temperatures of the prototype and model are less than 1%. To conclude this section, scaling up the prototype does not The slight disparities are due to mesh differences as well as change the physics of the problem. Thus, experiments are truncation and round off errors. Thus, these results demonstrate performed on a phantom identical to the scaled-up model of that the prototype can be scaled up without affecting the the scala tympani (3:1) for validating the computational heat physics of the problem. transfer model. This is discussed in the next section. The scale analysis also enable comparing the significance of natural convection with respect to conduction heat transfer IV. VALIDATION OF THE COM PUTATIONAL HEAT in the scala tympani. Based on preliminary simulations in the TRANSFER M ODEL WITH EXPERIMENTAL DATA scala tympani with an inserted implant electrode array, it is found that the first term on the left-hand side of (8) is on the The validation experiments are performed using a phantom order of one, the second term is on the order of -7, whereas the model of the scala tympani with a scale of 3:1. The details of term on the right-hand side is on the order of four. This means the phantom geometry are provided in [33], and the phantom 5 TABLE I T HERMOPHYSICAL PROPERTIES OF THE SUBSTANCES USED IN THE SIM ULATIONS. Substance Density Thermal conductivity Heat capacity kg W J [ ] k [ ] c [ ] 3 p m mK kgK Perilymph (Water) 992.20 [38] 0.625 [38] 4176.6 [38] Resistive wire (Ni-Cr80) 8400 [39] 11.3 [39] 450 [39] a a a Phantom material 1125.2 0.25 1610 b b b Magnet 430 8.1 7500 c c c Electrode array 19400 28 127.7 Blood 1050 [18] 0.52 [18] 3840 [18] Bone 1908 [18] 0.32 [18] 1313 [18] Provided by the manufacturer (Realize Inc.) Provided by the manufacturer (SUPERMAGNETMAN). Calculated based on the information provided by MED-EL. Fig. 5. Dimensionless temperature as a function of the dimensionless time (Fourier number) at the reference point for the phantom prototype and model. Fig. 3. Dimensionless temperature and velocity as a function of the dimen- sionless distance from the origin for the prototype and model at (x=0, y=0, z=0). Fig. 6. Dimensionless temperature as a function of the dimensionless location along the cut line defined in fig. 4(d) for the phantom prototype and model. the transient temperature along the scala tympani. A data logger (CAMPBELL SCIENTIFIC INC. CR5000) is used to record the temperature data. A water bath (VWR 1167 Heated Refrigerated Circulating Water Bath Polyscience) provides a constant and uniform temperature at the phantom bound- aries. The heat source is made of a resistance wire (Omega- NIC80-010-062) connected to two copper wires (Copper Wire 30AWG) and is controlled via a power supply (E3616A 60W Power Supply, 35V, 1.7A). All instruments are calibrated prior to each experiment based on their respective user manuals. A schematic and photographs of the experimental setup are Fig. 4. Scala tympani phantom: (a) Adiabatic boundary conditions are applied to all exterior walls. (b) Reference point where the transient temperature is shown in Fig. 7. calculated (prototype: x = 0 mm, y = 0 mm, z = -2/3 mm, model: x = 0 mm, Experimental data are used for validating the computational y = 0 mm, z = -2 mm). (c) Cut plane crossing the reference point. (d) Cut line heat transfer model. For that purpose, the 3D geometry of the on the cut plane where the spatial distribution of temperature is calculated. phantom has been imported in COMSOL Multiphysics. The thermocouples and their associated insertion holes have been STL files are available in [40]. The phantom is equipped removed in the numerical model to reduce the computational with 12 thermocouples (CHAL-015-BW-Omega) to measure load. This simplification does not affect the results in a 6 = 1000 is determined by fitting experimental data with m K numerical simulations. This value falls within the range of possible heat transfer coefficients for forced convection in the water bath, which is between 500 and 3,000 [41], [42]. m K Note that natural convection in the scala tympani is neglected. A sample validation result for thermocouple 2 is plotted in Fig. 8. The rate of temperature change is almost zero for t < 50 s because heat has not penetrated far enough into the phantom (i.e., thermocouple 2 does not yet feel the boundary condition). After 50 s, the heat from the water bath penetrates far enough in the phantom to increase the temperature at the thermocouple 2 location. The temperature increase continues at a lower rate after 350 s and approaches 22.4 C at 500 s. Due to the low heat transfer rate, it takes longer than 500 s for all points within the phantom to reach thermal equilibrium with the water at 25 C. The validation plots are similar for the other 11 thermo- couples. An error analysis has been conducted based on the ASME V&V 20 2009 Standard for Verification and Validation in Computational Fluid Dynamics and Heat Transfer [43]. To quantify the simulation error, the absolute difference between the average temperature recorded in seven experiments and the Fig. 7. Schematic (panel (a)) and photographs (panel (b)) of the experimental calculated temperature for each instant is calculated. Then, the setup. maximum difference between the numerical and experimental results for each thermocouple is divided by the maximum perceptible manner owing to the small size of the thermo- temperature difference, which is 25 C. Using this technique, couples. Based on simulations, it was mentioned in section it is found that the average error for all 12 thermocouples III that natural convection has a negligible impact on the is less than 5%. This difference is due to numerical errors, thermal management of the cochlea. This is explored further and the uncertainties in the input properties and the initial in this section by comparing the experimental data against the temperature. In addition, the short time required for transfer- numerical results generated by considering only conduction ring the phantom from the ice bath to the warm water bath within the scala tympani. is not taken into account in the numerical model, which may Preliminary simulations revealed that 5,178,756 mesh el- cause minor errors. Here, given the good agreement between ements and a time step of 0.1 s were sufficient to obtain experimental and numerical results, it is concluded that the converged results. Two scenarios are tested hereafter, namely model supports the assumption that natural convection within heat transfer within the scala tympani phantom subjected to a the scala tympani is negligible. sudden change in its thermal environment, and heat transfer within the scala tympani phantom subjected to localized B. Scala tympani subjected to localized heating heating via a small heat source. Localized heating is provided by a heat source connected to the power supply and embedded close to the apex of the scala A. Scala-tympani subjected to a sudden change in its thermal tympani (i.e., narrowest part of the scala tympani at the top of environment the phantom) as shown in Fig. 9. The scala tympani canal is In these experiments, the phantom is filled with distilled filled with distilled water and is immersed in the water bath. water and its two outlets are plugged with rubber. The phantom The scala tympani canal is plugged to avoid mixing the water is placed in a mixture of ice and water to reach a steady-state inside the bath with the distilled water within the phantom. temperature corresponding to the melting point of water in The temperature of the water bath is set to 11 C, and the the Salt Lake City geographical area, which is  0.004 C. temperature readings are monitored until all thermocouples Meanwhile, the water bath is heated to 25 C. Once steady- show steady-state temperatures equal to 11 C. The voltage state temperatures are reached, the phantom is subjected and current supplied to the heat source are controlled by the to a sudden change in its thermal environment by quickly power supply. As the heater circuit is in series, the input transferring it to the water bath until it reaches a temperature current is the same in the whole circuit. The total resistance of 25 C. The transient temperature of each thermocouple is of the circuit is slightly higher than the resistance of the recorded during this process. resistive wire (Ni-Cr80) located inside the phantom. The total For the numerical simulations, the entire system is assumed resistance of the circuit includes the resistance of the copper to be at an initial temperature of 0 C. A convective boundary wires in addition to the resistance of the Ni-Cr80 wire. The condition at the phantom boundary with a constant ambient resistance of the Ni-Cr80 wire (R = 2.1 ) and the input temperature of 25 C is modeled. A heat transfer coefficient h current (I = 0.3 A) are used to calculate the input power. 7 Fig. 9. 3D model of the scala tympani subjected to localized heating with a heat source used in the simulations for comparison against experimental data. The top surface of the phantom is removed for better visualization of the scala tympani. Fig. 8. Temperature as a function of time when the scala tympani is subjected to a sudden change in its thermal environment. Experimental data are compared against numerical simulations. Transient temperatures of all 12 thermocouples are recorded for validating the computational heat transfer model. For the simulations, the initial temperature of the entire system is set to 11 C. A convective boundary condition with an ambient temperature of 11 C and a heat transfer coefficient of h = 1000 , as determined in the experiments without a heat m K source, is used. The heat source provides a power of 189 mW. It is found that the average difference between the model and experiment data, considering all 12 thermocouples, is less than 6% by using the technique described in section IV.A Fig. 10. Temperature as a function of time when the scala tympani is subjected and repeating each experiment five times. The error is due to to localized heating. Experimental data are compared against numerical the uncertainties in the input data, including the heat transfer simulations. coefficient and the magnitude of the heat source input power, in addition to the previously mentioned numerical errors. Fig. V. MAXIMUM SAFE INPUT POW ER TO AVOID LOCALIZED 10 shows sample validation results for thermocouples 2, 9, and HYPERTHERMIA the thermocouple embedded within the heat source. Similar trends are obtained for all thermocouples. The temperature of The validated 3D heat transfer model of the scala tympani is the heat source increases within 200 s to 27 C, after which applied hereafter to calculate the maximum safe input power to steady-state conditions are reached where the heat removal rate avoid localized hyperthermia when detaching the magnet from balances the input power. As expected, the rate of temperature the implant electrode array. The maximum safe input power change at other locations within the scala tympani decreases is determined using the thermal dose or cumulative equivalent with increasing distance from the heat source. For example, minutes (CEM). The CEM evaluates thermal damage of tissues the temperature recorded by thermocouple 2 (located 9.6 mm as a function of temperature and length of exposure to an away from heat source) changes minimally during the transient elevated temperature. The thermal threshold of different types process, while the temperature measured by thermocouple 9 of tissues are in the range of 40 C to 55 C [44], [45]. The (located 6.8 mm away from the heat source) increases from death rate of most types of tissues increases at 43 C, such 11 C to 15 C. that this temperature is selected as a reference for calculating In conclusion, a good agreement between experimental data the CEM. The safe length of exposure to 43 C for coclear and numerical results has also been obtained with localized tissues is 114 s [46]. This means that cochlear tissues are not heating. As such, the 3D heat transfer model of the scala damaged if they are exposed to a temperature of 43 C during tympani can be applied with confidence to predict the maxi- 114 s after which the temperature suddenly drops to the body mum safe input power to avoid localized hyperthermia when core temperature. During the actual process, tissues are heated detaching the magnet after magnetic cochlear implant surgery. from the body core temperature to 43 C, or tissues may even 8 be heated to temperatures higher than 43 C. After turning off the heat source, cochlear tissues cool down gradually, and, as a result, they stay at a temperature higher than the body core temperature for a certain amount of time after detaching the magnet. This cooling period is taken into account in the calculation of thermal dose which is given by [44], [45]: t=t (43T ) CEM = t B (12) 43 n t=0 Fig. 11. 3D model of the scala tympani with inserted magnet and cochlear where t is the total length of exposure to all temperatures, implant electrode array. t is the length of exposure to the specific temperature T , n n Scala vestibuli and B is a constant equal to 0.25 for temperatures lower than 43 C and 0.5 for temperatures higher than 43 C. In Bone ( Low temperature ) the simulations, the magnet is heated at discrete input power densities, for a fixed period of time, then the heating process is stopped and the simulation is continued until the maximum temperature in the cochlea, implant electrode array and magnet is less than 37.05 C. The CEM for the whole process, Membrane ( High temperature ) including heating and cooling, is calculated. The maximum Scala tympani Scala media power density providing a CEM smaller than 114 s but larger than 113 s is selected as the maximum safe input power Fig. 12. Characteristic length used to estimate the heat transfer coefficient density to detach the magnet. from the scala tympani to the scala vestibuli and scala media. The geometry of the scala tympani numerical model used in section IV is modified for calculating the maximum safe input power. Simulations are performed in the previous section on a by a semi-elliptical cross-section (see Fig 12), and it is model involving a hollow scala tympani within a surrounding assumed that the membrane is at 43 C and the bony wall is at material in order to reproduce as faithfully as possible the 37 C. The Rayleigh number Ra is less than 1000 within the 3D-printed phantom. Modeling a finite surrounding material scala tympani, scala media, and scala vestibuli based on a 6C is not required when dealing with the actual situation where temperature difference, and a characteristic length of 1.64 mm. the implant electrode array is inserted in the scala tympani. Bouras et al. [47] calculated Nu for a semi-elliptical geometry Therefore, to find the maximum safe input power to detach and showed that for Ra less than 5000, Nu is equal to 1. The Nuk hL fluid the magnet, the finite material surrounding the scala tympani Nusselt number Nu is defined as , thus h = . k L fluid c and the resistive wire that was used as the heat source are The characteristic length is usually defined as the thickness of removed from the model used in section IV. Instead, the the boundary layer. In this study, the length a as defined in Fig. implant electrode array and the magnet are added to the model 12 is assumed to be the characteristic length. The diameter of (see Fig. 11). The magnet is modeled as a cylinder with a the scala media is negligible in comparison to the diameter of diameter of 0.5 mm and length of 1 mm. The electrode array the scala vestibuli. The diameter of the scala vestibuli varies is modeled as a 31-mm-long, 0.2-mm-diameter cylinder. The from 1.64 mm at the base of the cochlea to 0.81 mm at thermophysical properties of the implant electrode array and the apex of the cochlea [48]. As such, this diameter range the magnet are listed in Table I. In Fig. 11, the membrane is used to estimate the heat transfer coefficient. The minimum between the scala tympani and scala media is defined as and maximum heat transfer coefficients are 385 and 775 m K boundary 1, while boundaries 2 and 3 are respectively the bony . These two extreme values are tested in scenarios 4 and m K boundary of the scala tympani and the insertion opening. The 5 (see Table II). scala media has a relatively small thickness in comparison to The maximum safe input power density to avoid local- the scala vestibuli and scala tympani, such that it is assumed ized hyperthermia when detaching the magnet is calculated that the scala media and scala vestibuli form a single region. for eleven discrete heating periods and for the five sets of Boundary 1 is assumed to be either isothermal at the body boundary conditions listed in Table II. The results are plotted core temperature of 37 C or adiabatic, boundary 2 is assumed in Fig. 13. As expected, the maximum safe input power to be isothermal, adiabatic, or subjected to convection, while density increases by reducing the heating period. Among the boundary 3 is always isothermal at the body core temperature. five sets of boundary conditions considered, scenario 2 where The five different sets of boundary conditions considered in boundaries 1 and 2 are adiabatic is the most conservative. For a the simulations are listed in Table II. given heating period, scenarios 4 and 5 involving convection at It is assumed that heat is transferred from the scala tympani boundary 1 are the closest to the adiabatic boundary condition. Therefore, the adiabatic boundary condition (scenario 2) has to the scala media and scala vestibuli by convection. The possible values for the heat transfer coefficient at boundary the highest safety factor and ensures that tissues will not suffer 1 are estimated by evaluating the magnitude of the Nusselt thermal damage during the magnet detachment process. number Nu. The scala media and scala vestibuli are described As discussed in section III, assuming that the magnet is 9 TABLE II power to avoid localized hyperthermia when detaching the S ETS OF BOUNDARY CONDITIONS USED TO CALCULATE THE M AXIMUM magnet from the implant electrode array. Five different com- SAFE INPUT POWER DENSITY. binations of boundary conditions and eleven discrete heating Boundary 1 Boundary 2 Boundary 3 periods were considered in the simulations. It was shown Scenario 1 Isothermal (37 C) Isothermal (37 C) Isothermal (37 C) Scenario 2 Adiabatic Adiabatic Isothermal (37 C) that the power required to release the magnet attached to Scenario 3 Isothermal (37 C) Adiabatic Isothermal (37 C) W   the implant electrode array by 1 mm of paraffin was less Scenario 4 Convection (h = 385 ; T =37 C ) Adiabatic Isothermal (37 C) m K Scenario 5 Convection (h = 775 , T =37 C ) Adiabatic Isothermal (37 C) than the maximum safe input power calculated for the most m K conservative case where all boundaries, except the insertion opening, are adiabatic. The results of this study will enable designing a thermally safe procedure for magnetic cochlear implant surgery. In addition, the validated 3D heat transfer model is not limited to cochlear implant surgery and can be applied to other thermal management problems related to hearing science. ACKNOWLEDGMENTS Research reported in this publication was supported by the National Institute on Deafness and Other Communication Disorders of the National Institutes of Health under Award Number R01DC013168. The content is solely the responsibil- ity of the authors and does not necessarily represent the official views of the National Institutes of Health. We acknowledge the Center for High Performance Computing at the University of Utah for their support and resources, and in particular Dr. Martin Cuma for his assistance in facilitating the numerical computations. We would also like to show our gratitude to Fig. 13. Maximum safe input power density to avoid hyperthermia when the MED-EL company for sharing the information about detaching the magnet as a function of the heating time and sets of boundary conditions defined in Table II. their electrode arrays. We would like to thank Dr. Lisandro Leon and Matt Cavilla for designing and fabricating the scala tympani phantom and Travis Morrison for helping to set up attached to the implant electrode array by 1 mm of paraffin, the experiments. a power of 21 mW during 10 s is necessary to melt the paraffin 6 3 at 43 C (21  10 W/m ). These calculations assume that REFERENCES the paraffin boundaries are adiabatic and that all the heat is transferred from the magnet to the paraffin. The maximum [1] J. R. Clark et al., “Magnetic Guidance of Cochlear Implants: Proof-of- Concept and Initial Feasibility Study,” in Journal of Medical Devices,, calculated safe input power when boundaries 1 and 2 are vol. 6, no. 3, 2012 , pp. 035002. Available: doi = 10.1115/1.4007099 assumed to be adiabatic is 43.9 mW during 10 s (43.9 [2] L. Leon, F. M. Warren, J. J. Abbott, “An In-Vitro Insertion-Force 6 3 10 W/m ), which is higher than the energy required to melt Study of Magnetically Guided Lateral-Wall Cochlear-Implant Electrode Arrays,” in Otol Neurotol,, vol. 39, no. 2, 2018 , pp. e63-e73. Available: the paraffin. doi:10.1097/MAO.0000000000001647 To conclude, the magnet can be detached from the implant [3] T. L. Bruns et al., “ Magnetically Steered Robotic Insertion of Cochlear- electrode array without causing localized hyperthermia. For Implant Electrode Arrays: System Integration and First-In-Cadaver Re- sults,” in IEEE Robotics and Automation Letters,, vol. 5, no. 2, 2020 , the most efficient and safest result, it is better to apply higher pp. 2240-2247. Available: doi: 10.1109/LRA.2020.2970978 input power density in a shorter period of time and use a grade [4] F. Esmailie, M. Francoeur, and T. Ameel, “Heat transfer of paraffin that has a melting temperature lower than 43 C but analysis in an uncoiled model of the cochlea during magnetic cochlear implant surgery,” in International Journal of Heat higher than 38 C. and Mass Transfer,, vol. 154, 2020, pp. 119683. Available: https://doi.org/10.1016/j.ijheatmasstransfer.2020.119683 [5] F. Esmailie, M. S. Cavilla,and T. A. Ameel, “A Thermal Transient VI. CONCLUSIONS Model of Heat Transfer Within an Omnimagnet,” in Proceedings of the ASME 2017 International Mechanical Engineering Congress A 3D heat transfer model of the scala tympani has been and Exposition. Volume 8: Heat Transfer and Thermal Engineer- validated using a phantom subjected to a sudden change in its ing,, Tampa, Florida, USA, 2017, pp. V008T10A046. Available: thermal environment and to localized heating. Comparison of https://doi.org/10.1115/IMECE2017-72120 [6] R. T. Richardson et al., “Challenges for the application of op- measured and predicted transient temperatures revealed that tical stimulation in the cochlea for the study and treatment of the average errors were less than 5% and 6% without and hearing loss,” in Expert Opinion on Biological Therapy,, vol. with a heat source, respectively. This work also confirmed 17, no. 2, Taylor & Francis, 2017, pp. 213–223. Available: https://doi.org/10.1080/14712598.2017.1271870 that natural convection heat transfer has a negligible impact [7] A. C. Thompson et al., “Infrared nerve stimulation: modelling of photon on the thermal management of the scala tympani with respect transport and heat conduction,” in Proceedings Volume 8579, Optical to conduction heat transfer. Finally, the validated heat transfer Interactions with Tissue and Cells XXIV,, vol. 8579, SPIE, 2013, pp. 45 model was applied to determine the maximum safe input – 51. Available: https://doi.org/10.1117/12.2003477 10 [8] A. C. Thompson et al., “‘Modeling of the temporal effects Cochlea,” in IEEE Transactions on Biomedical Engineering,, vol. 63, no. of heating during infrared neural stimulation,” in Journal of 1, 2016, pp. 188–198. Available: DOI: 10.1109/TBME.2015.2480601 Biomedical Optics,, vol. 18, no. 3, 2013, pp. 035004. Available: [28] N. S. Tachos et al., “Middle and inner ear modelling: From microCT https://doi.org/10.1117/1.JBO.18.3.035004 images to 3D reconstruction and coupling of models,” in 38th Annual [9] M. G. Shapiro et al., “Infrared light excites cells by changing their International Conference of the IEEE Engineering in Medicine and electrical capacitance,” in Nature Communications 3,, vol. 736, 2012. Biology Society (EMBC), Orlando, FL, USA, 2016, pp. 5961–5964. Available: https://doi.org/10.1038/ncomms1742 Available: DOI: 10.1109/EMBC.2016.7592086 [29] J. Wysocki, “Dimensions of the human vestibular and tympanic scalae,” [10] N. Xia et al., “Pressure in the Cochlea During Infrared Irradiation,” in in Hearing Research,, vol. 135, no. 1 2, 1999, pp. 39 – 46. Available: IEEE Transactions on Biomedical Engineering,, vol. 65, no. 7, 2018, https://doi.org/10.1016/S0378-5955(99)00088-X pp. 1575–1584. Available: https://doi.org/10.1109/TBME.2016.2636149 [30] S. Raufer, J. J. Guinan, H. H. Nakajima, “Cochlear partition anatomy [11] S. Kodali, S.A. Harvey,and T.E. Prieto, “Thermal effects of and motion in humans differ from the classic view of mammals,” in Pro- laser stapedectomy in an animal model: CO versus KTP,” ceedings of the National Academy of Sciences,, vol. 116, no. 28, 2019, in Laryngoscope,, vol. 107, 1997 , pp. 14451450. Available: pp. 113977-13982. Available: https://doi.org/10.1073/pnas.1900787116 doi:10.1097/00005537-199711000-00005 [31] S. K. Yoo et al., “Three-dimensional geometric modeling of the cochlea [12] M. Szymaski,K. Morshed, and R. Mills, “Experimental study on heat using helico-spiral approximation,” in IEEE Transactions on Biomed- transmission to the vestibule during CO laser use in revision stapes ical Engineering,, vol. 47, no. 10, 2000, pp. 1392-1402. Available: surgery,” in J Laryngol Otol,, vol. 121, no. 1, 2007 , pp. 5-8. Available: DOI: 10.1109/10.871413 https://doi.org/10.1017/S0022215106002672 [32] S. Biedron et al., “The Internal Dimensions of the Cochlear Scalae [13] A. J. Baertschi, R.N. Johnson, and G.R. Hanna, “A theoretical and With Special Reference to Cochlear Electrode Insertion Trauma,” in experimental determination of vestibular dynamics in caloric stim- Otology & Neurotology,, vol. 31, no. 5, 2010, pp. 731 – 737. Available: ulation,” in Biol Cybern,, vol. 20, 1975 , pp. 175186. Available: doi: 10.1097/MAO.0b013e3181d27b5e https://doi.org/10.1007/BF00342638 [33] L. Leon et al., “Scala-Tympani phantom with cochleostomy and round- [14] T. CAWTHORNE, and W. A, COBB, “Temperature changes in the window openings for cochlear-implant insertion experiments,” in Journal perilymph space in response to caloric stimulation in man,” in Acta of Medical Devices,, vol. 8, no. 2, 2014, pp. 041010 – 041020. Available: Otolaryngol,, vol. 44, no. 5 6, 1954 , pp. 580-588. Available: https://doi.org/10.1115/1.4027617 https://doi.org/10.3109/00016485409127670 [34] H. H. Pennes, “Analysis of Tissue and Arterial Blood Temperatures in [15] R. L. McIntosh, V. Anderson, R. and J. McKenzie, “A numerical evalua- the Resting Human Forearm,” in J. Appl. Physiol,, vol. 1, 1948, pp. 5 tion of SAR distribution and temperature changes around a metallic plate – 34. Available: DOI: 10.1152/jappl.1998.85.1.5 in the head of a RF exposed worker,” in Bioelectromagnetics,, vol. 26, [35] P. Pelliccia et al., “Cochlea size variability and implications in clinical no. 5, 2005 , pp. 377388. Available: https://doi.org/10.1002/bem.20112 practice,” in Acta Otorhinolaryngol Ital,, vol. 34, no. 1, 2014, pp. 42–49. [16] P. Bernardi et al., “Specific absorption rate and temperature increases in [36] N. Gerber et al., “A multiscale imaging and modelling dataset of the head of a cellular-phone user,” in IEEE Transactions on Microwave the human inner ear,” in Sci Data,, vol. 4, 2017, 170132. Available: Theory and Techniques,, vol. 48, no. 7, 2000 , pp. 1118-1126. Available: https://doi.org/10.1038/sdata.2017.132 doi: 10.1109/22.848494 [37] J. H. Kwonet al., “The thickness and texture of temporal bone in [17] R. L. McIntosh et al., “Assessment of SAR and thermal changes brain CT predict acoustic window failure of transcranial Doppler,” near a cochlear implant system for mobile phone type expo- in Journal of Neuroimaging,, vol. 16, 2006 , pp. 347-352. Available: sures,” in Bioelectromagnetics,, vol. 29, 2008 , pp. 71-80. Available: doi:10.1111/j.1552-6569.2006.00064.x doi:10.1002/bem.20364 [38] COMSOL Multiphysics, Application Programming [18] I. Tamames et al., “Theoretical Evaluation and Experimental Guide, COMSOL Multiphysics v. 5.4, 2019. Available: Validation of Localized Therapeutic Hypothermia Application https://cdn.comsol.com/doc/5.5/ApplicationProgrammingGuide.pdf to Preserve Residual Hearing After Cochlear Implantation,” [39] Nickel-Chromium Alloys (Ni-Cr), Reade International Corp., 2020. in Ear Hear,, vol. 39, no. 4, 2018, pp. 712–719. Available: Available: https://www.reade.com/products/nickel-chromium-alloys-nicr https://doi.org/10.1097/AUD.0000000000000529 [40] Robotic Assistance for Cochlear Implants, 2020. Available: [19] E. A. Dugan et al., “Therapeutic hypothermia reduces cortical in- https://www.telerobotics.utah.edu/index.php/Research/CochlearImplants flammation associated with Utah array implants,” in Journal of [41] HEAT TRANSFER COEFFICIENT, 2011. Available: Neural Engineering,, vol. 17, no. 2, 2020, pp. 026035. Available: DOI:10.1615/AtoZ.h.heat transfer coefficient https://iopscience.iop.org/article/10.1088/1741-2552/ab85d2 [42] CONVECTIVE HEAT TRANSFER, 2011. Available: [20] R. Liljemalm, T. Nyberg, H. von Holst, “Heating during infrared neural DOI:10.1615/AtoZ.c.convective heat transfer stimulation,” in Lasers Surg Med,, vol. 45, no. 7, 2013 , pp. 469481. [43] ASME Standard for Verification and Validation in Computational Fluid Available: https://doi.org/10.1002/lsm.22158 Dynamics and Heat Transfer V V 20 - 2009. [21] J. R. Clark, , F. M. Warren and J. J. Abbott, , “A Scalable Model for [44] P. S. Yarmolenko et al., “Thresholds for thermal damage to normal Human Scala-Tympani Phantoms,” in J. Med. Device.,, vol. 5, no. 1, tissues: an update,” in International journal of hyperthermia : the 2011, pp. 14501 – 14505. official journal of European Society for Hyperthermic Oncology, North [22] G. Tognola et al., “Experimental Measurements of Potentials Generated American Hyperthermia Group, vol. 27, no. 4, 2011, pp. 320–343. by the Electrodes of a Cochlear Implant in a Phantom,” in 1th Mediter- Available: https://doi.org/10.3109/02656736.2010.534527 ranean Conference on Medical and Biomedical Engineering and Com- [45] G. C. Van Rhoon et al., “CEM thermal dose thresholds: a poten- puting 2007. IFMBE Proceedings Berlin, Heidelberg, Germany, 2007, tial guide for magnetic resonance radiofrequency exposure levels?,” pp. 390–392. Available: https://doi.org/10.1007/978-3- 540-73044- 6 99 in European Radiology, vol. 23, 2013, pp. 2215–2227. Available: [23] O. Majdani et al., “Artifacts Caused by Cochlear Implants with Non- https://doi.org/10.1007/s00330-013-2825-y Removable Magnets in 3T MRI: Phantom and Cadaveric Studies,” [46] N. Yoshida, A. Kristiansen, and M. C. Liberman, “Heat stress and in Eur Arch Otorhinolaryngol,, vol. 266, 2009, pp. 1885. Available: protection from permanent acoustic injury in mice,” in Journal of https://doi.org/10.1007/s00405-009-0994-8 Neuroscience,, vol. 19, no. 22, 1999, pp. 10116–10124. Available: [24] R. Z. Gan, B. P. Reeves, and X. Wang, “Modeling of Sound Transmis- https://doi.org/10.1523/JNEUROSCI.19-22-10116.1999 sion from Ear Canal to Cochlea,” in Ann. Biomed. Eng.,, vol. 35, 2007, [47] A. Bouras et al., “Laminar Natural Convection Study in a Hori- pp. 2180 – 2195. Available: https://doi.org/10.1007/s10439-007-9366-y zontal Half-Elliptical Enclosure Using Heater on Horizontal Wall,” [25] R. Hayden et al., “Virtual Labyrinth Model of Vestibular Af- in Journal of Advanced Research in Fluid Mechanics and Ther- ferent Excitation via Implanted Electrodes: Validation and Ap- mal Sciences,, vol. 53, no. 2, 2019, pp. 223–233. Available: plication to Design of a Multichannel Vestibular Prosthesis,” in http://dspace.univ-msila.dz:8080//xmlui/handle/123456789/15253 Exp. brain Res.,, vol. 210, 2011, pp. 623 – 640. Available: [48] B. Slavomir et al., “The Internal Dimensions of the Cochlear Scalae https://doi.org/10.1007/s00221-011-2599-x With Special Reference to Cochlear Electrode Insertion Trauma,” in [26] C. A. Shera, J. J. Guinan, and A. J. Oxenham, “Otoacoustic Estimation Otology & Neurotology,, vol. 31, no. 5, 2010, pp. 731–737. Available: of Cochlear Tuning: Validation in the Chinchilla,” in Journal of the doi: 10.1097/MAO.0b013e3181d27b5e Association for Research in Otolaryngology,, vol. 11, 2010, pp. 343 – 365. Available: https://doi.org/10.1007/s10162-010-0217-4 [27] P. Wong et al., “Development and Validation of a High-Fidelity Finite- Element Model of Monopolar Stimulation in the Implanted Guinea Pig http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Physics arXiv (Cornell University)

Experimental validation of a three-dimensional heat transfer model within the scala tympani with application to magnetic cochlear implant surgery

Esmailie, Fateme; Francoeur, Mathieu; Ameel, Tim
Physics , Volume 2021 (2009) – Sep 15, 2020
Download PDF
10 pages

Loading next page...
 
/lp/arxiv-cornell-university/experimental-validation-of-a-three-dimensional-heat-transfer-model-MLN2QxyxLk

References (45)

ISSN
0018-9294
eISSN
ARCH-3341
DOI
10.1109/TBME.2021.3055976
Publisher site
See Article on Publisher Site

Abstract

Experimental validation of a three-dimensional heat transfer model within the scala tympani with application to magnetic cochlear implant surgery Fateme Esmailie, Mathieu Francoeur, Tim Ameel Abstract—Magnetic guidance of cochlear implants is a promis- a magnet in the cochlea may cause medical complications if ing technique to reduce the risk of physical trauma during the patient is exposed to a strong external magnetic such as surgery. In this approach, a magnet attached to the tip of the in magnetic resonance imaging. The removal can be accom- implant electrode array is guided within the scala tympani using a plished by melting the substance bonding the implant electrode magnetic field. After surgery, the magnet must be detached from array and the magnet via localized heating. Consequently, the implant electrode array via localized heating and removed from the scala tympani which may cause thermal trauma. the magnet detachment process may cause localized hyper- Objectives: The objective of this work is to experimentally thermia within the scala tympani, which is one of the more validate a three-dimensional (3D) heat transfer model of the serious impediments preventing the establishment of magnetic scala tympani which will enable accurate predictions of the cochlear implant surgery [4], [5]. Localized hyperthermia in maximum safe input power to avoid localized hyperthermia when the ear, which is not unique to magnetic cochlear implant detaching the magnet from the implant electrode array. Methods: Experiments are designed using a rigorous scale analysis and surgery, arises in a variety of applications such as infrared performed by measuring transient temperatures in a 3D-printed neural stimulation implants [6] – [10], stapedectomy [11], [12], scala tympani phantom subjected to a sudden change in its caloric test [13], [14], and radio-frequency radiation devices thermal environment and localized heating via a small heat such as cellular phones [15] – [17]. In addition, studies have source. Results: The measured and predicted temperatures are shown that localized hypothermia can preserve tissues after in good agreement with an error less than 6%. Conclusions: The validated 3D heat transfer model of the scala tympani is cochlear implantation [18], [19]. finally applied to evaluate the maximum safe input power to Despite the importance of localized hyperthermia and hy- avoid localized hyperthermia when detaching the magnet. For the pothermia in hearing research, validated three-dimensional most conservative case where all boundaries except the insertion (3D) heat transfer computational models within cochlear opening are adiabatic, the power required to release the magnet canals are scarce. Thompson et al. [6] – [8] and Liljemalm [20] attached to the implant electrode array by 1 mm of paraffin is approximately half of the predicted maximum safe input power. developed a heat transfer model to evaluate the thermal impact Significance: This work will enable the design of a thermally safe of infrared neural stimulation implants. In this technique, the magnetic cochlear implant surgery procedure. implant focuses infrared radiation for stimulating modulus Index Terms—Cochlear implant, Magnetic guidance, Scala nerves, which are a bundle of nerves located in the middle of tympani, Heat transfer simulations, Heat transfer experiments the bony cochlea. They validated their numerical results using the experimental data of Shapiro et al. [9], who conducted in-vitro experiments by stimulating an oocyte. Thompson et al. [6] – [8] and Liljemalm [20] analyzed the feasibility of infrared neural stimulation implants, but the actual geometry I. I NTRODUCTION of the cochlea was not considered in the validation phase of Magnetic guidance of cochlear implants is a promising these works. Similar to magnetic cochlear implant surgery, surgical technique that is expected to mitigate physical trauma infrared neural stimulation implant surgery involves localized arising with manual insertion of implants [1]– [3]. In this heating. However, neither the heat source nor the targeted technique, a magnetic field guides the implant electrode array tissues are similar to the magnet detachment process. The within the scala tympani, one of the cochlear canals, using a impact of therapeutic hypothermia in the cochlea was studied small magnet attached to the implant tip. After surgery, the by Tamames et al. [18] using a COMSOL Multiphysics model magnet must be detached from the implant electrode array validated with experimental data. The focus of their work was and removed from the scala tympani, since the presence of on the preservation of cochlear tissues via hypothermia and the design of a cooling device. Whereas this work is an excellent Submission date: 09/01/2020 Research reported in this publication was supported by the National Institute example of a validated heat transfer model within the cochlea, on Deafness and Other Communication Disorders of the National Institutes the application and physics are different from the magnetic of Health under Award Number R01DC013168. cochlear implant surgery problem. F. Esmailie is with Department of Mechanical Engineering, University of Utah, Salt Lake City, Utah 84112, USA. The objective of this paper is, therefore, to validate a M. Francoeur is with Department of Mechanical Engineering, University 3D heat transfer computational model of the scala tympani, of Utah, Salt Lake City, Utah 84112, USA . which will enable quantification of the risks of localized T. Ameel is with Department of Mechanical Engineering, University of Utah, Salt Lake City, Utah 84112, USA (e-mail: ameel@mech.utah.edu). hyperthermia during magnetic cochlear implant surgery. In arXiv:2009.07199v1 [physics.app-ph] 15 Sep 2020 2 Fig. 1. Cutaway view of the cochlea. The cochlea includes three canals, Fig. 2. 3D drawing of the scala tympani phantom. namely the scala tympani, the scala vestibuli, and the scala media. The cochlear implant electrode array is magnetically-guided in the scala tympani. physical scala tympani model used in the experiments will be hearing research, new surgical methods or treatments are tested referred to as the phantom. The phantom, consisting of a hol- in phantoms [21] – [23], cadavers [23], [24], or animals low scala tympani canal, was 3D-printed by Realize Inc. with [25] – [27]. Here, transient temperatures measured in a 3D- a transparent material called Somos WaterShed XC 11122. The phantom includes twelve ducts enabling the insertion of printed scala tympani phantom subjected to a sudden change thermocouples for measuring transient temperatures along the in its thermal environment and localized heating are used for scala tympani. A separate insertion opening included in the validating the computational model. As the numerical and phantom facilitates implantation of an implant (see Fig. 2). experimental results are in good agreement, the model is Transient temperatures are measured for two distinct cases, applied to determine the maximum safe input power to avoid namely when the phantom is subjected to a sudden change in localized hyperthermia when detaching the magnet. its thermal environment, and when there is localized heating The balance of the paper is organized as follows. Section within the phantom. The experimental data are then compared II is devoted to a description of the problem. Subsequently, the design of the experiment used to validate the heat transfer against numerical predictions obtained with the computational model and the associated scale analysis are discussed. Tran- heat transfer model having a geometry identical to that of the sient temperatures measured in a 3D-printed scala tympani phantom. phantom subjected to a sudden change in its thermal environ- A deterrent to thermocouple temperature measurement at ment and localized heating are compared against numerical small scales is the possible influence they may have on the predictions in section IV. Finally, the validated heat transfer measurement. If the size of the thermocouples is of the same model is used to predict the maximum input power to avoid order of magnitude as the phantom, the thermocouples may hyperthermia in the scala tympani for various boundary con- affect the temperature distribution and fluid flow within the ditions and heating durations. scala tympani canal, and may act as heat sinks. To avoid these potential issues, a scaled-up phantom is used in the II. DESCRIPTION OF THE PROBLEM experiments. This is discussed next. The inner ear consists of the cochlea and the organ of III. EXPERIM ENTAL DESIGN AND SCALE ANALYSIS balance [28]. The cochlea is a spiral bony organ with two and a half turns consisting of three canals, namely the scala vestibuli, The impact of phantom scale on the physics of the problem the scala media and the scala tympani (see Fig. 1) [29] – and the significance of each term in the energy and momentum [31]. The scala tympani and scala vestibuli are filled with a balance equations describing fluid and heat transport is studied dilute saline fluid called perilymph, while the scala media is via a scale analysis. The scale analysis is a well-established filled with endolymph which is also a dilute saline fluid [29] method in thermal-fluid science that enables eliminating negli- – [31]. Both the perilymph and endolymph are characterized gible physical phenomena in a process, and permits designing by thermophysical properties identical to that of water. The physically meaningful experiments. cochlear implant is magnetically guided in the scala tympani. The scale analysis requires non-dimensionalizing the energy and momentum balance equations. A general form of the Physical models of the scala tympani have been previously energy balance equation is given by [34]: constructed [32], [33]. In this work, the physical model devel- oped by Lisandro et al. [33] (see Fig. 2) is used for validating @T c + c urT +r (krT ) +r q + p p the computational heat transfer model within the scala tym- @t (1) pani. To avoid confusion with the computational model, the c ! (T T ) = q + q bl p;bl bl bl met 3 where q is the radiative heat flux vector, and q and q are heat transfer (diffusion). The energy balance in the perilymph met respectively the volumetric heat sources due to metabolic heat does not include the input power density, but accounts for heat generation and the magnet. Blood perfusion (last term on the transfer by conduction and convection. The energy balance left-hand side of (1) in the cochlea absorbs heat and contributes equation in the perilymph region is: to energy dissipation. The blood perfusion rate in the cochlea is @T ml less than 1 [8]. In infrared neural stimulation, the energy c + c urT +r (krT ) = 0 (3) ming p p @t dissipated by perfusion in the cochlea is negligible compared to the magnitude of the heat source [7], [9], [18], [19], [8]. where the second term on the left-hand side is the heat trans- For example, Thompson et al. [7], [8] discussed all possible port by convection. To calculate this term, the velocity vector biological cooling methods in the cochlea (perfusion, cochlear fluid flow, evaporation of water from tissues), and concluded should be simultaneously computed by solving the momentum balance (i.e., Navier-Stokes) equation in the perilymph: that these mechanisms are much lower than the magnitude of the laser heat source power (6.25 mW) used for infrared neural @u stimulation implants. Here, an input power of 21 mW applied +  (ur) u = r pI + g (4) @t for 10 s is necessary to detach the magnet by assuming that the magnet is attached to the implant electrode array by 1 where the first term on the left-hand side is the acceleration mm of paraffin having a melting point of 43 C. This is more term and the second term represents advection. The first term than three times larger than the power reported by Thompson on the right-hand side is the pressure force and the second et al. [7], [8]. The conclusions of Thompson et al. [7], [8] term represents the body force (gravity) acting in the negative may be applicable for magnetic cochlear implant surgery, but z direction. Note that the external magnetic field guiding heat dissipation by blood perfusion should also be compared the magnet through the scala tympani is turned off after the to other heat dissipation mechanisms in the scala tympani, and surgery, such that magnetohydrodynamic effects do not come not solely to the external input power. into the picture. According to [7], the heat removal rate per unit volume The momentum and energy balance equations are non- by perfusion in the ear for a temperature difference of 3 C ml dimensionalized hereafter for the scale analysis. The dimen- and a perfusion rate of 1 is  0.03 mW. For magnetic ming sionless Cartesian coordinates are defined as: cochlear implant surgery, the temperature difference may be up to 6 C, such that the perfusion heat removal rate is estimated x y z at 0.06 mW (assuming a cochlear volume of 185.4 mm X = ; Y = ; Z = (5) kW L L L c c c [35] and a heat removal rate of 0.33 ). In contrast, heat dissipation through the cochlear temporal bone characterized where L is a characteristic length. Similarly, the Cartesian by a surface area of 250 mm [36] and a thickness of 4 mm components of the velocity vector are non-dimensionalized as [37] is approximately 120 mW (assuming a cochlear volume follows: 3 kW of 185.4 mm [35] and a heat removal rate of 647.2 as u u u estimated by Fouriers law). As a result, the blood perfusion x y z U = ; U = ; U = ; (6) x y z term in (1) is negligible with respect to heat dissipation by u u u 0 0 0 conduction through the temporal bone. p where u (= g L (T T )) is derived by assuming that 0 c max i The metabolic heat generation rate per unit volume (q ) in met kW the gravitational and viscous forces are in equilibrium. The a cochlea is approximately 1.1 [18]. Under the assumption TT dimensionless temperature is defined as  = , where that the entire cochlea is filled with blood, which is an T T max i T is initial temperature, and the dimensionless time, also overestimation, the metabolic heat generation is  0.2 mW. called the Fourier number, is given by Fo = . This is two orders of magnitude smaller than the magnitude L Substituting the dimensionless parameters into (2) to (4), of the heat source (21 mW), such that the metabolic heat the dimensionless energy balance in the solid magnet region generation term can be neglected in (1). is: Heat transport by radiation is estimated at 0.01 mW using 2 2 2 @ @  @  @ the Stefan-Boltzmann law, which is three orders of magnitude = + + + Q (7) 2 2 2 @Fo @X @Y @Z smaller than the heat source. The contribution to heat transfer by radiation is therefore neglected in (1). qL After neglecting blood perfusion, metabolic heat generation, where Q = is the dimensionless input power. The V k(T T ) max i and radiation heat transfer, the simplified energy balance dimensionless energy balance in the perilymph is written as: equation applied to the solid magnet region where heat is @ @ @ @ 0:5 transferred solely by conduction is given by: + PrGr U + U + U = x y z @Fo @X @Y @Z (8) @T 2 2 2 @  @  @ c +r (krT ) = q (2) + + @t 2 2 2 @X @Y @Z g (T T )L s 1 The first term on the left-hand side of (2) represents thermal where Pr = and Gr = are the Prandtl energy storage, whereas the second term is the conduction and Grashoff numbers, respectively. Finally, the dimensionless 4 momentum balance equations in the perilymph along each that natural convection heat transfer (second term on the left- Cartesian coordinate are given by: hand side of (8)) is negligible in comparison to conduction (right-hand side of (8)). This result is in agreement with our 1 @U @U @U @U x x x x 0:5 previous study performed on an uncoiled cochlea model [4], + Gr U + U + U = x y z Pr @Fo @X @Y @Z where it was shown that natural convection has a negligible (9) 2 2 2 @ U @ U @ U x x x impact on the thermal management of the scala tympani during + + 2 2 2 @X @Y @Z magnetic cochlear implant surgery. As such, heat transfer by natural convection is neglected in the remainder of this paper. Note that the accuracy of this assumption is later confirmed 1 @U @U @U @U y y y y 0:5 + Gr U + U + U = x y z with experimental data. Pr @Fo @X @Y @Z (10) In addition to the two-concentric cylinder problem, the 2 2 2 @ U @ U @ U y y y accuracy of the scale analysis for a prototype and model (3:1) + + 2 2 2 @X @Y @Z of the scala tympani is also analyzed. The scale analysis is performed based on the geometry and thermophysical proper- 1 @U @U @U @U ties of the scala tympani phantom discussed in section II. A z z z z 0:5 + Gr U + U + U = x y z spiral-shaped heat source is implanted in the narrowest region Pr @Fo @X @Y @Z (11) 2 2 2 of the phantom, while it is assumed that all boundaries of the @ U @ U @ U z z z 0:5 + + + Gr phantom are adiabatic and that the initial temperature is 0 C. 2 2 2 @X @Y @Z The thermophysical properties of perilymph (same as water), Based on (7) to (11), the scale analysis implies that if the the phantom prototype material (Somos WaterShed XC11122), dimensionless parameters Fo; Pr; Gr; and Q are equal for the and the heat source (Ni-Cr80 wire) are listed in Table I. actual physical model (prototype) and the scaled-up physical The geometry of the scala tympani prototype is illustrated model (simply called model hereafter), then the dimensionless in Fig. 4 (a). For equivalent values of Q, the power inputs temperature  and dimensionless velocities U ; U ; U will be for the phantom prototype and model are 0.9 W and 0.1 W, x y z equal for the prototype and the model. respectively. Temperatures are calculated at a reference point The scale analysis is first verified by solving a simple (prototype: x = 0 mm, y = 0 mm, z = -2/3 mm, model: x = heat transfer problem between two concentric cylinders. The 0 mm, y = 0 mm, z = -2 mm) and a cut line on the cut plane inner cylinder is heated via constant heat rates of 0.01 W crossing the reference point (see Figs. 4, (b), (c) and (d)). (model) and 0.04 W (prototype). The boundary of the outer The transient dimensionless temperature at the reference point cylinder is adiabatic, and the gap region between the two is plotted in Fig. 5. As expected, the temperature increases cylinders is filled with water. The thermophysical properties with increasing time when the heat source is turned on. The required for the simulations are listed in Table I. The initial temperature gradient increases after some time and, as a result, temperature of the entire system is 20 C. For the prototype, the temperature increase rate is enhanced. This trend is the the diameter and length of the inner cylinder are 1 mm and 5 same for both the phantom prototype and model with less than mm, respectively. The outer cylinder has a diameter of 2 mm 5% difference between individual data. The differences are due and a length of 5 mm. Here, the model is two times larger to the disparate meshes for the prototype and model as well as than the prototype (see Fig. 3). The characteristic length is truncation and round off errors. The dimensionless temperature defined as the gap distance between the two cylinders. The along the cut line in the phantom prototype is plotted in velocity at the surface of the inner and outer cylinders is zero Fig. 6. The temperature is maximum near the heat source. because of the no-slip boundary condition. Consequently, the As the distance with respect to the heat source increases, dimensionless velocity is zero, and reaches a maximum value the temperature gradient decreases and, consequently, the heat between the two cylinders. The dimensionless velocities for transfer rate drops. The spatial temperature distribution trend the prototype and model are in a good agreement with less is identical for the phantom prototype and model with less than 1% difference. The temperature at the surface of the than 5% discrepancy at a few locations. Again, the differences heated inner cylinder is a maximum and decreases towards the are due to different mesh as well as truncation and round off outer cylinder. Again, differences between the dimensionless errors. temperatures of the prototype and model are less than 1%. To conclude this section, scaling up the prototype does not The slight disparities are due to mesh differences as well as change the physics of the problem. Thus, experiments are truncation and round off errors. Thus, these results demonstrate performed on a phantom identical to the scaled-up model of that the prototype can be scaled up without affecting the the scala tympani (3:1) for validating the computational heat physics of the problem. transfer model. This is discussed in the next section. The scale analysis also enable comparing the significance of natural convection with respect to conduction heat transfer IV. VALIDATION OF THE COM PUTATIONAL HEAT in the scala tympani. Based on preliminary simulations in the TRANSFER M ODEL WITH EXPERIMENTAL DATA scala tympani with an inserted implant electrode array, it is found that the first term on the left-hand side of (8) is on the The validation experiments are performed using a phantom order of one, the second term is on the order of -7, whereas the model of the scala tympani with a scale of 3:1. The details of term on the right-hand side is on the order of four. This means the phantom geometry are provided in [33], and the phantom 5 TABLE I T HERMOPHYSICAL PROPERTIES OF THE SUBSTANCES USED IN THE SIM ULATIONS. Substance Density Thermal conductivity Heat capacity kg W J [ ] k [ ] c [ ] 3 p m mK kgK Perilymph (Water) 992.20 [38] 0.625 [38] 4176.6 [38] Resistive wire (Ni-Cr80) 8400 [39] 11.3 [39] 450 [39] a a a Phantom material 1125.2 0.25 1610 b b b Magnet 430 8.1 7500 c c c Electrode array 19400 28 127.7 Blood 1050 [18] 0.52 [18] 3840 [18] Bone 1908 [18] 0.32 [18] 1313 [18] Provided by the manufacturer (Realize Inc.) Provided by the manufacturer (SUPERMAGNETMAN). Calculated based on the information provided by MED-EL. Fig. 5. Dimensionless temperature as a function of the dimensionless time (Fourier number) at the reference point for the phantom prototype and model. Fig. 3. Dimensionless temperature and velocity as a function of the dimen- sionless distance from the origin for the prototype and model at (x=0, y=0, z=0). Fig. 6. Dimensionless temperature as a function of the dimensionless location along the cut line defined in fig. 4(d) for the phantom prototype and model. the transient temperature along the scala tympani. A data logger (CAMPBELL SCIENTIFIC INC. CR5000) is used to record the temperature data. A water bath (VWR 1167 Heated Refrigerated Circulating Water Bath Polyscience) provides a constant and uniform temperature at the phantom bound- aries. The heat source is made of a resistance wire (Omega- NIC80-010-062) connected to two copper wires (Copper Wire 30AWG) and is controlled via a power supply (E3616A 60W Power Supply, 35V, 1.7A). All instruments are calibrated prior to each experiment based on their respective user manuals. A schematic and photographs of the experimental setup are Fig. 4. Scala tympani phantom: (a) Adiabatic boundary conditions are applied to all exterior walls. (b) Reference point where the transient temperature is shown in Fig. 7. calculated (prototype: x = 0 mm, y = 0 mm, z = -2/3 mm, model: x = 0 mm, Experimental data are used for validating the computational y = 0 mm, z = -2 mm). (c) Cut plane crossing the reference point. (d) Cut line heat transfer model. For that purpose, the 3D geometry of the on the cut plane where the spatial distribution of temperature is calculated. phantom has been imported in COMSOL Multiphysics. The thermocouples and their associated insertion holes have been STL files are available in [40]. The phantom is equipped removed in the numerical model to reduce the computational with 12 thermocouples (CHAL-015-BW-Omega) to measure load. This simplification does not affect the results in a 6 = 1000 is determined by fitting experimental data with m K numerical simulations. This value falls within the range of possible heat transfer coefficients for forced convection in the water bath, which is between 500 and 3,000 [41], [42]. m K Note that natural convection in the scala tympani is neglected. A sample validation result for thermocouple 2 is plotted in Fig. 8. The rate of temperature change is almost zero for t < 50 s because heat has not penetrated far enough into the phantom (i.e., thermocouple 2 does not yet feel the boundary condition). After 50 s, the heat from the water bath penetrates far enough in the phantom to increase the temperature at the thermocouple 2 location. The temperature increase continues at a lower rate after 350 s and approaches 22.4 C at 500 s. Due to the low heat transfer rate, it takes longer than 500 s for all points within the phantom to reach thermal equilibrium with the water at 25 C. The validation plots are similar for the other 11 thermo- couples. An error analysis has been conducted based on the ASME V&V 20 2009 Standard for Verification and Validation in Computational Fluid Dynamics and Heat Transfer [43]. To quantify the simulation error, the absolute difference between the average temperature recorded in seven experiments and the Fig. 7. Schematic (panel (a)) and photographs (panel (b)) of the experimental calculated temperature for each instant is calculated. Then, the setup. maximum difference between the numerical and experimental results for each thermocouple is divided by the maximum perceptible manner owing to the small size of the thermo- temperature difference, which is 25 C. Using this technique, couples. Based on simulations, it was mentioned in section it is found that the average error for all 12 thermocouples III that natural convection has a negligible impact on the is less than 5%. This difference is due to numerical errors, thermal management of the cochlea. This is explored further and the uncertainties in the input properties and the initial in this section by comparing the experimental data against the temperature. In addition, the short time required for transfer- numerical results generated by considering only conduction ring the phantom from the ice bath to the warm water bath within the scala tympani. is not taken into account in the numerical model, which may Preliminary simulations revealed that 5,178,756 mesh el- cause minor errors. Here, given the good agreement between ements and a time step of 0.1 s were sufficient to obtain experimental and numerical results, it is concluded that the converged results. Two scenarios are tested hereafter, namely model supports the assumption that natural convection within heat transfer within the scala tympani phantom subjected to a the scala tympani is negligible. sudden change in its thermal environment, and heat transfer within the scala tympani phantom subjected to localized B. Scala tympani subjected to localized heating heating via a small heat source. Localized heating is provided by a heat source connected to the power supply and embedded close to the apex of the scala A. Scala-tympani subjected to a sudden change in its thermal tympani (i.e., narrowest part of the scala tympani at the top of environment the phantom) as shown in Fig. 9. The scala tympani canal is In these experiments, the phantom is filled with distilled filled with distilled water and is immersed in the water bath. water and its two outlets are plugged with rubber. The phantom The scala tympani canal is plugged to avoid mixing the water is placed in a mixture of ice and water to reach a steady-state inside the bath with the distilled water within the phantom. temperature corresponding to the melting point of water in The temperature of the water bath is set to 11 C, and the the Salt Lake City geographical area, which is  0.004 C. temperature readings are monitored until all thermocouples Meanwhile, the water bath is heated to 25 C. Once steady- show steady-state temperatures equal to 11 C. The voltage state temperatures are reached, the phantom is subjected and current supplied to the heat source are controlled by the to a sudden change in its thermal environment by quickly power supply. As the heater circuit is in series, the input transferring it to the water bath until it reaches a temperature current is the same in the whole circuit. The total resistance of 25 C. The transient temperature of each thermocouple is of the circuit is slightly higher than the resistance of the recorded during this process. resistive wire (Ni-Cr80) located inside the phantom. The total For the numerical simulations, the entire system is assumed resistance of the circuit includes the resistance of the copper to be at an initial temperature of 0 C. A convective boundary wires in addition to the resistance of the Ni-Cr80 wire. The condition at the phantom boundary with a constant ambient resistance of the Ni-Cr80 wire (R = 2.1 ) and the input temperature of 25 C is modeled. A heat transfer coefficient h current (I = 0.3 A) are used to calculate the input power. 7 Fig. 9. 3D model of the scala tympani subjected to localized heating with a heat source used in the simulations for comparison against experimental data. The top surface of the phantom is removed for better visualization of the scala tympani. Fig. 8. Temperature as a function of time when the scala tympani is subjected to a sudden change in its thermal environment. Experimental data are compared against numerical simulations. Transient temperatures of all 12 thermocouples are recorded for validating the computational heat transfer model. For the simulations, the initial temperature of the entire system is set to 11 C. A convective boundary condition with an ambient temperature of 11 C and a heat transfer coefficient of h = 1000 , as determined in the experiments without a heat m K source, is used. The heat source provides a power of 189 mW. It is found that the average difference between the model and experiment data, considering all 12 thermocouples, is less than 6% by using the technique described in section IV.A Fig. 10. Temperature as a function of time when the scala tympani is subjected and repeating each experiment five times. The error is due to to localized heating. Experimental data are compared against numerical the uncertainties in the input data, including the heat transfer simulations. coefficient and the magnitude of the heat source input power, in addition to the previously mentioned numerical errors. Fig. V. MAXIMUM SAFE INPUT POW ER TO AVOID LOCALIZED 10 shows sample validation results for thermocouples 2, 9, and HYPERTHERMIA the thermocouple embedded within the heat source. Similar trends are obtained for all thermocouples. The temperature of The validated 3D heat transfer model of the scala tympani is the heat source increases within 200 s to 27 C, after which applied hereafter to calculate the maximum safe input power to steady-state conditions are reached where the heat removal rate avoid localized hyperthermia when detaching the magnet from balances the input power. As expected, the rate of temperature the implant electrode array. The maximum safe input power change at other locations within the scala tympani decreases is determined using the thermal dose or cumulative equivalent with increasing distance from the heat source. For example, minutes (CEM). The CEM evaluates thermal damage of tissues the temperature recorded by thermocouple 2 (located 9.6 mm as a function of temperature and length of exposure to an away from heat source) changes minimally during the transient elevated temperature. The thermal threshold of different types process, while the temperature measured by thermocouple 9 of tissues are in the range of 40 C to 55 C [44], [45]. The (located 6.8 mm away from the heat source) increases from death rate of most types of tissues increases at 43 C, such 11 C to 15 C. that this temperature is selected as a reference for calculating In conclusion, a good agreement between experimental data the CEM. The safe length of exposure to 43 C for coclear and numerical results has also been obtained with localized tissues is 114 s [46]. This means that cochlear tissues are not heating. As such, the 3D heat transfer model of the scala damaged if they are exposed to a temperature of 43 C during tympani can be applied with confidence to predict the maxi- 114 s after which the temperature suddenly drops to the body mum safe input power to avoid localized hyperthermia when core temperature. During the actual process, tissues are heated detaching the magnet after magnetic cochlear implant surgery. from the body core temperature to 43 C, or tissues may even 8 be heated to temperatures higher than 43 C. After turning off the heat source, cochlear tissues cool down gradually, and, as a result, they stay at a temperature higher than the body core temperature for a certain amount of time after detaching the magnet. This cooling period is taken into account in the calculation of thermal dose which is given by [44], [45]: t=t (43T ) CEM = t B (12) 43 n t=0 Fig. 11. 3D model of the scala tympani with inserted magnet and cochlear where t is the total length of exposure to all temperatures, implant electrode array. t is the length of exposure to the specific temperature T , n n Scala vestibuli and B is a constant equal to 0.25 for temperatures lower than 43 C and 0.5 for temperatures higher than 43 C. In Bone ( Low temperature ) the simulations, the magnet is heated at discrete input power densities, for a fixed period of time, then the heating process is stopped and the simulation is continued until the maximum temperature in the cochlea, implant electrode array and magnet is less than 37.05 C. The CEM for the whole process, Membrane ( High temperature ) including heating and cooling, is calculated. The maximum Scala tympani Scala media power density providing a CEM smaller than 114 s but larger than 113 s is selected as the maximum safe input power Fig. 12. Characteristic length used to estimate the heat transfer coefficient density to detach the magnet. from the scala tympani to the scala vestibuli and scala media. The geometry of the scala tympani numerical model used in section IV is modified for calculating the maximum safe input power. Simulations are performed in the previous section on a by a semi-elliptical cross-section (see Fig 12), and it is model involving a hollow scala tympani within a surrounding assumed that the membrane is at 43 C and the bony wall is at material in order to reproduce as faithfully as possible the 37 C. The Rayleigh number Ra is less than 1000 within the 3D-printed phantom. Modeling a finite surrounding material scala tympani, scala media, and scala vestibuli based on a 6C is not required when dealing with the actual situation where temperature difference, and a characteristic length of 1.64 mm. the implant electrode array is inserted in the scala tympani. Bouras et al. [47] calculated Nu for a semi-elliptical geometry Therefore, to find the maximum safe input power to detach and showed that for Ra less than 5000, Nu is equal to 1. The Nuk hL fluid the magnet, the finite material surrounding the scala tympani Nusselt number Nu is defined as , thus h = . k L fluid c and the resistive wire that was used as the heat source are The characteristic length is usually defined as the thickness of removed from the model used in section IV. Instead, the the boundary layer. In this study, the length a as defined in Fig. implant electrode array and the magnet are added to the model 12 is assumed to be the characteristic length. The diameter of (see Fig. 11). The magnet is modeled as a cylinder with a the scala media is negligible in comparison to the diameter of diameter of 0.5 mm and length of 1 mm. The electrode array the scala vestibuli. The diameter of the scala vestibuli varies is modeled as a 31-mm-long, 0.2-mm-diameter cylinder. The from 1.64 mm at the base of the cochlea to 0.81 mm at thermophysical properties of the implant electrode array and the apex of the cochlea [48]. As such, this diameter range the magnet are listed in Table I. In Fig. 11, the membrane is used to estimate the heat transfer coefficient. The minimum between the scala tympani and scala media is defined as and maximum heat transfer coefficients are 385 and 775 m K boundary 1, while boundaries 2 and 3 are respectively the bony . These two extreme values are tested in scenarios 4 and m K boundary of the scala tympani and the insertion opening. The 5 (see Table II). scala media has a relatively small thickness in comparison to The maximum safe input power density to avoid local- the scala vestibuli and scala tympani, such that it is assumed ized hyperthermia when detaching the magnet is calculated that the scala media and scala vestibuli form a single region. for eleven discrete heating periods and for the five sets of Boundary 1 is assumed to be either isothermal at the body boundary conditions listed in Table II. The results are plotted core temperature of 37 C or adiabatic, boundary 2 is assumed in Fig. 13. As expected, the maximum safe input power to be isothermal, adiabatic, or subjected to convection, while density increases by reducing the heating period. Among the boundary 3 is always isothermal at the body core temperature. five sets of boundary conditions considered, scenario 2 where The five different sets of boundary conditions considered in boundaries 1 and 2 are adiabatic is the most conservative. For a the simulations are listed in Table II. given heating period, scenarios 4 and 5 involving convection at It is assumed that heat is transferred from the scala tympani boundary 1 are the closest to the adiabatic boundary condition. Therefore, the adiabatic boundary condition (scenario 2) has to the scala media and scala vestibuli by convection. The possible values for the heat transfer coefficient at boundary the highest safety factor and ensures that tissues will not suffer 1 are estimated by evaluating the magnitude of the Nusselt thermal damage during the magnet detachment process. number Nu. The scala media and scala vestibuli are described As discussed in section III, assuming that the magnet is 9 TABLE II power to avoid localized hyperthermia when detaching the S ETS OF BOUNDARY CONDITIONS USED TO CALCULATE THE M AXIMUM magnet from the implant electrode array. Five different com- SAFE INPUT POWER DENSITY. binations of boundary conditions and eleven discrete heating Boundary 1 Boundary 2 Boundary 3 periods were considered in the simulations. It was shown Scenario 1 Isothermal (37 C) Isothermal (37 C) Isothermal (37 C) Scenario 2 Adiabatic Adiabatic Isothermal (37 C) that the power required to release the magnet attached to Scenario 3 Isothermal (37 C) Adiabatic Isothermal (37 C) W   the implant electrode array by 1 mm of paraffin was less Scenario 4 Convection (h = 385 ; T =37 C ) Adiabatic Isothermal (37 C) m K Scenario 5 Convection (h = 775 , T =37 C ) Adiabatic Isothermal (37 C) than the maximum safe input power calculated for the most m K conservative case where all boundaries, except the insertion opening, are adiabatic. The results of this study will enable designing a thermally safe procedure for magnetic cochlear implant surgery. In addition, the validated 3D heat transfer model is not limited to cochlear implant surgery and can be applied to other thermal management problems related to hearing science. ACKNOWLEDGMENTS Research reported in this publication was supported by the National Institute on Deafness and Other Communication Disorders of the National Institutes of Health under Award Number R01DC013168. The content is solely the responsibil- ity of the authors and does not necessarily represent the official views of the National Institutes of Health. We acknowledge the Center for High Performance Computing at the University of Utah for their support and resources, and in particular Dr. Martin Cuma for his assistance in facilitating the numerical computations. We would also like to show our gratitude to Fig. 13. Maximum safe input power density to avoid hyperthermia when the MED-EL company for sharing the information about detaching the magnet as a function of the heating time and sets of boundary conditions defined in Table II. their electrode arrays. We would like to thank Dr. Lisandro Leon and Matt Cavilla for designing and fabricating the scala tympani phantom and Travis Morrison for helping to set up attached to the implant electrode array by 1 mm of paraffin, the experiments. a power of 21 mW during 10 s is necessary to melt the paraffin 6 3 at 43 C (21  10 W/m ). These calculations assume that REFERENCES the paraffin boundaries are adiabatic and that all the heat is transferred from the magnet to the paraffin. The maximum [1] J. R. Clark et al., “Magnetic Guidance of Cochlear Implants: Proof-of- Concept and Initial Feasibility Study,” in Journal of Medical Devices,, calculated safe input power when boundaries 1 and 2 are vol. 6, no. 3, 2012 , pp. 035002. Available: doi = 10.1115/1.4007099 assumed to be adiabatic is 43.9 mW during 10 s (43.9 [2] L. Leon, F. M. Warren, J. J. Abbott, “An In-Vitro Insertion-Force 6 3 10 W/m ), which is higher than the energy required to melt Study of Magnetically Guided Lateral-Wall Cochlear-Implant Electrode Arrays,” in Otol Neurotol,, vol. 39, no. 2, 2018 , pp. e63-e73. Available: the paraffin. doi:10.1097/MAO.0000000000001647 To conclude, the magnet can be detached from the implant [3] T. L. Bruns et al., “ Magnetically Steered Robotic Insertion of Cochlear- electrode array without causing localized hyperthermia. For Implant Electrode Arrays: System Integration and First-In-Cadaver Re- sults,” in IEEE Robotics and Automation Letters,, vol. 5, no. 2, 2020 , the most efficient and safest result, it is better to apply higher pp. 2240-2247. Available: doi: 10.1109/LRA.2020.2970978 input power density in a shorter period of time and use a grade [4] F. Esmailie, M. Francoeur, and T. Ameel, “Heat transfer of paraffin that has a melting temperature lower than 43 C but analysis in an uncoiled model of the cochlea during magnetic cochlear implant surgery,” in International Journal of Heat higher than 38 C. and Mass Transfer,, vol. 154, 2020, pp. 119683. Available: https://doi.org/10.1016/j.ijheatmasstransfer.2020.119683 [5] F. Esmailie, M. S. Cavilla,and T. A. Ameel, “A Thermal Transient VI. CONCLUSIONS Model of Heat Transfer Within an Omnimagnet,” in Proceedings of the ASME 2017 International Mechanical Engineering Congress A 3D heat transfer model of the scala tympani has been and Exposition. Volume 8: Heat Transfer and Thermal Engineer- validated using a phantom subjected to a sudden change in its ing,, Tampa, Florida, USA, 2017, pp. V008T10A046. Available: thermal environment and to localized heating. Comparison of https://doi.org/10.1115/IMECE2017-72120 [6] R. T. Richardson et al., “Challenges for the application of op- measured and predicted transient temperatures revealed that tical stimulation in the cochlea for the study and treatment of the average errors were less than 5% and 6% without and hearing loss,” in Expert Opinion on Biological Therapy,, vol. with a heat source, respectively. This work also confirmed 17, no. 2, Taylor & Francis, 2017, pp. 213–223. Available: https://doi.org/10.1080/14712598.2017.1271870 that natural convection heat transfer has a negligible impact [7] A. C. Thompson et al., “Infrared nerve stimulation: modelling of photon on the thermal management of the scala tympani with respect transport and heat conduction,” in Proceedings Volume 8579, Optical to conduction heat transfer. Finally, the validated heat transfer Interactions with Tissue and Cells XXIV,, vol. 8579, SPIE, 2013, pp. 45 model was applied to determine the maximum safe input – 51. Available: https://doi.org/10.1117/12.2003477 10 [8] A. C. Thompson et al., “‘Modeling of the temporal effects Cochlea,” in IEEE Transactions on Biomedical Engineering,, vol. 63, no. of heating during infrared neural stimulation,” in Journal of 1, 2016, pp. 188–198. Available: DOI: 10.1109/TBME.2015.2480601 Biomedical Optics,, vol. 18, no. 3, 2013, pp. 035004. Available: [28] N. S. Tachos et al., “Middle and inner ear modelling: From microCT https://doi.org/10.1117/1.JBO.18.3.035004 images to 3D reconstruction and coupling of models,” in 38th Annual [9] M. G. Shapiro et al., “Infrared light excites cells by changing their International Conference of the IEEE Engineering in Medicine and electrical capacitance,” in Nature Communications 3,, vol. 736, 2012. Biology Society (EMBC), Orlando, FL, USA, 2016, pp. 5961–5964. Available: https://doi.org/10.1038/ncomms1742 Available: DOI: 10.1109/EMBC.2016.7592086 [29] J. Wysocki, “Dimensions of the human vestibular and tympanic scalae,” [10] N. Xia et al., “Pressure in the Cochlea During Infrared Irradiation,” in in Hearing Research,, vol. 135, no. 1 2, 1999, pp. 39 – 46. Available: IEEE Transactions on Biomedical Engineering,, vol. 65, no. 7, 2018, https://doi.org/10.1016/S0378-5955(99)00088-X pp. 1575–1584. Available: https://doi.org/10.1109/TBME.2016.2636149 [30] S. Raufer, J. J. Guinan, H. H. Nakajima, “Cochlear partition anatomy [11] S. Kodali, S.A. Harvey,and T.E. Prieto, “Thermal effects of and motion in humans differ from the classic view of mammals,” in Pro- laser stapedectomy in an animal model: CO versus KTP,” ceedings of the National Academy of Sciences,, vol. 116, no. 28, 2019, in Laryngoscope,, vol. 107, 1997 , pp. 14451450. Available: pp. 113977-13982. Available: https://doi.org/10.1073/pnas.1900787116 doi:10.1097/00005537-199711000-00005 [31] S. K. Yoo et al., “Three-dimensional geometric modeling of the cochlea [12] M. Szymaski,K. Morshed, and R. Mills, “Experimental study on heat using helico-spiral approximation,” in IEEE Transactions on Biomed- transmission to the vestibule during CO laser use in revision stapes ical Engineering,, vol. 47, no. 10, 2000, pp. 1392-1402. Available: surgery,” in J Laryngol Otol,, vol. 121, no. 1, 2007 , pp. 5-8. Available: DOI: 10.1109/10.871413 https://doi.org/10.1017/S0022215106002672 [32] S. Biedron et al., “The Internal Dimensions of the Cochlear Scalae [13] A. J. Baertschi, R.N. Johnson, and G.R. Hanna, “A theoretical and With Special Reference to Cochlear Electrode Insertion Trauma,” in experimental determination of vestibular dynamics in caloric stim- Otology & Neurotology,, vol. 31, no. 5, 2010, pp. 731 – 737. Available: ulation,” in Biol Cybern,, vol. 20, 1975 , pp. 175186. Available: doi: 10.1097/MAO.0b013e3181d27b5e https://doi.org/10.1007/BF00342638 [33] L. Leon et al., “Scala-Tympani phantom with cochleostomy and round- [14] T. CAWTHORNE, and W. A, COBB, “Temperature changes in the window openings for cochlear-implant insertion experiments,” in Journal perilymph space in response to caloric stimulation in man,” in Acta of Medical Devices,, vol. 8, no. 2, 2014, pp. 041010 – 041020. Available: Otolaryngol,, vol. 44, no. 5 6, 1954 , pp. 580-588. Available: https://doi.org/10.1115/1.4027617 https://doi.org/10.3109/00016485409127670 [34] H. H. Pennes, “Analysis of Tissue and Arterial Blood Temperatures in [15] R. L. McIntosh, V. Anderson, R. and J. McKenzie, “A numerical evalua- the Resting Human Forearm,” in J. Appl. Physiol,, vol. 1, 1948, pp. 5 tion of SAR distribution and temperature changes around a metallic plate – 34. Available: DOI: 10.1152/jappl.1998.85.1.5 in the head of a RF exposed worker,” in Bioelectromagnetics,, vol. 26, [35] P. Pelliccia et al., “Cochlea size variability and implications in clinical no. 5, 2005 , pp. 377388. Available: https://doi.org/10.1002/bem.20112 practice,” in Acta Otorhinolaryngol Ital,, vol. 34, no. 1, 2014, pp. 42–49. [16] P. Bernardi et al., “Specific absorption rate and temperature increases in [36] N. Gerber et al., “A multiscale imaging and modelling dataset of the head of a cellular-phone user,” in IEEE Transactions on Microwave the human inner ear,” in Sci Data,, vol. 4, 2017, 170132. Available: Theory and Techniques,, vol. 48, no. 7, 2000 , pp. 1118-1126. Available: https://doi.org/10.1038/sdata.2017.132 doi: 10.1109/22.848494 [37] J. H. Kwonet al., “The thickness and texture of temporal bone in [17] R. L. McIntosh et al., “Assessment of SAR and thermal changes brain CT predict acoustic window failure of transcranial Doppler,” near a cochlear implant system for mobile phone type expo- in Journal of Neuroimaging,, vol. 16, 2006 , pp. 347-352. Available: sures,” in Bioelectromagnetics,, vol. 29, 2008 , pp. 71-80. Available: doi:10.1111/j.1552-6569.2006.00064.x doi:10.1002/bem.20364 [38] COMSOL Multiphysics, Application Programming [18] I. Tamames et al., “Theoretical Evaluation and Experimental Guide, COMSOL Multiphysics v. 5.4, 2019. Available: Validation of Localized Therapeutic Hypothermia Application https://cdn.comsol.com/doc/5.5/ApplicationProgrammingGuide.pdf to Preserve Residual Hearing After Cochlear Implantation,” [39] Nickel-Chromium Alloys (Ni-Cr), Reade International Corp., 2020. in Ear Hear,, vol. 39, no. 4, 2018, pp. 712–719. Available: Available: https://www.reade.com/products/nickel-chromium-alloys-nicr https://doi.org/10.1097/AUD.0000000000000529 [40] Robotic Assistance for Cochlear Implants, 2020. Available: [19] E. A. Dugan et al., “Therapeutic hypothermia reduces cortical in- https://www.telerobotics.utah.edu/index.php/Research/CochlearImplants flammation associated with Utah array implants,” in Journal of [41] HEAT TRANSFER COEFFICIENT, 2011. Available: Neural Engineering,, vol. 17, no. 2, 2020, pp. 026035. Available: DOI:10.1615/AtoZ.h.heat transfer coefficient https://iopscience.iop.org/article/10.1088/1741-2552/ab85d2 [42] CONVECTIVE HEAT TRANSFER, 2011. Available: [20] R. Liljemalm, T. Nyberg, H. von Holst, “Heating during infrared neural DOI:10.1615/AtoZ.c.convective heat transfer stimulation,” in Lasers Surg Med,, vol. 45, no. 7, 2013 , pp. 469481. [43] ASME Standard for Verification and Validation in Computational Fluid Available: https://doi.org/10.1002/lsm.22158 Dynamics and Heat Transfer V V 20 - 2009. [21] J. R. Clark, , F. M. Warren and J. J. Abbott, , “A Scalable Model for [44] P. S. Yarmolenko et al., “Thresholds for thermal damage to normal Human Scala-Tympani Phantoms,” in J. Med. Device.,, vol. 5, no. 1, tissues: an update,” in International journal of hyperthermia : the 2011, pp. 14501 – 14505. official journal of European Society for Hyperthermic Oncology, North [22] G. Tognola et al., “Experimental Measurements of Potentials Generated American Hyperthermia Group, vol. 27, no. 4, 2011, pp. 320–343. by the Electrodes of a Cochlear Implant in a Phantom,” in 1th Mediter- Available: https://doi.org/10.3109/02656736.2010.534527 ranean Conference on Medical and Biomedical Engineering and Com- [45] G. C. Van Rhoon et al., “CEM thermal dose thresholds: a poten- puting 2007. IFMBE Proceedings Berlin, Heidelberg, Germany, 2007, tial guide for magnetic resonance radiofrequency exposure levels?,” pp. 390–392. Available: https://doi.org/10.1007/978-3- 540-73044- 6 99 in European Radiology, vol. 23, 2013, pp. 2215–2227. Available: [23] O. Majdani et al., “Artifacts Caused by Cochlear Implants with Non- https://doi.org/10.1007/s00330-013-2825-y Removable Magnets in 3T MRI: Phantom and Cadaveric Studies,” [46] N. Yoshida, A. Kristiansen, and M. C. Liberman, “Heat stress and in Eur Arch Otorhinolaryngol,, vol. 266, 2009, pp. 1885. Available: protection from permanent acoustic injury in mice,” in Journal of https://doi.org/10.1007/s00405-009-0994-8 Neuroscience,, vol. 19, no. 22, 1999, pp. 10116–10124. Available: [24] R. Z. Gan, B. P. Reeves, and X. Wang, “Modeling of Sound Transmis- https://doi.org/10.1523/JNEUROSCI.19-22-10116.1999 sion from Ear Canal to Cochlea,” in Ann. Biomed. Eng.,, vol. 35, 2007, [47] A. Bouras et al., “Laminar Natural Convection Study in a Hori- pp. 2180 – 2195. Available: https://doi.org/10.1007/s10439-007-9366-y zontal Half-Elliptical Enclosure Using Heater on Horizontal Wall,” [25] R. Hayden et al., “Virtual Labyrinth Model of Vestibular Af- in Journal of Advanced Research in Fluid Mechanics and Ther- ferent Excitation via Implanted Electrodes: Validation and Ap- mal Sciences,, vol. 53, no. 2, 2019, pp. 223–233. Available: plication to Design of a Multichannel Vestibular Prosthesis,” in http://dspace.univ-msila.dz:8080//xmlui/handle/123456789/15253 Exp. brain Res.,, vol. 210, 2011, pp. 623 – 640. Available: [48] B. Slavomir et al., “The Internal Dimensions of the Cochlear Scalae https://doi.org/10.1007/s00221-011-2599-x With Special Reference to Cochlear Electrode Insertion Trauma,” in [26] C. A. Shera, J. J. Guinan, and A. J. Oxenham, “Otoacoustic Estimation Otology & Neurotology,, vol. 31, no. 5, 2010, pp. 731–737. Available: of Cochlear Tuning: Validation in the Chinchilla,” in Journal of the doi: 10.1097/MAO.0b013e3181d27b5e Association for Research in Otolaryngology,, vol. 11, 2010, pp. 343 – 365. Available: https://doi.org/10.1007/s10162-010-0217-4 [27] P. Wong et al., “Development and Validation of a High-Fidelity Finite- Element Model of Monopolar Stimulation in the Implanted Guinea Pig

Journal

Physics – arXiv (Cornell University)

Published: Sep 15, 2020

There are no references for this article.