The mechanics of the outer and middle ear have been the objective of much research and many publications. There is a common position regarding the main behavior of the system, especially at lower frequencies, however at higher frequencies there are still some controversies. At lower frequencies (at the range below 1–2 kHz), the classic piston-like motion model is widely accepted to described the mechanics of the sound transmission. Experimental and numerical research confirms this basic model.
Nevertheless, at frequencies above this range, different coupled mechanisms appear which are difficult to identify. Aspects as how the tympanic cavity (TC) couples to the tympanic membrane (TM), the influence of the ossicular system on the dynamic of the membrane or the effect of the external ear canal (EEC) over the eardrum motion and transfer functions are certainly unclear nowadays.
There are some common surgical interventions, as tymplanoplasties or bone substitution by prosthesis, which solve hearing problem but sometimes derive on an audition quality lost. For the patient, it becomes manifest in limitation on the speech intelligibility or difficulties to appreciate music. This is normally due to changes on the transmission of higher frequency sounds. A better understanding of the sound transmission mechanisms at these frequencies is the base for proper interventions or therapies. For instance, in the case of a tympanoplasty, the change on the membrane thickness change its behavior at higher frequencies.
The main objective of this paper is to find evidence of the TC role in the transfer function. This has a double benefit, on the one hand we obtain a better comprehension of the behavior of the system, on the other hand we can shed light on the interpretation of experimental work which commonly has to manipulate the TC in order to carry out the experiment.
There is a great deal of work published about the influence of the ear canal (EC), TM and ossicular chain (OC) in the auditory system (AS). Some papers show experimental measurements varying TC conditions in order to deduce the TC role in the AS. Otherwise, experimental difficulties limit their finding and it is not clear how the TC affects TM motion at high frequencies.
Regarding numerical model and particularly in finite element method (FEM), only a few papers have included TC in their Models [1–6], and only some of them [1–3] focused on the specific role that the EC, TM, and TC connection plays in the human AS. There is previous FEM work on TC influence on TM displacement (not in transfer functions) and on pressure gain produced by the EC [1], the main conclusion reported about the role of the middle ear cavities was that when cavities were opened the TM displacement was increased by a factor of two at low frequencies. The main restriction of the model presented by Koike et al. comes from the air model. The air was modeled as an elastic solid (as all other structures) with a very low Young modulus. There is a FEM paper [2] that analyzes the effect of mastoid cavity in EC pressures and umbo displacement (UD) (not in transfer function). The authors concluded that the pressure and UDs were slightly influenced by the status of the aditus (open or close). The aditus is a small connection between the TC and the mastoid cavities. Gan et al. have developed a relative complete human AS FEM [3–6], but only one of them [3] focused on the possible mechano-acoustic relation between EC, TM and TC, they presented in their study the influence of eardrum perforations would produce on the EC pressure. In other mammals like cats, there is FEM work presenting a model in which all cavities are modeled, and it is clearly visible and demonstrated that the coupling of the cavities of the middle ear to the eardrum causes a resonance around 5 kHz [7].
Regarding experimental researches, there is evidence that changing middle ear cavities produce changes in middle ear impedances [8], so it should affect the outer and middle ear transfer functions. Other works only shows results in a frequency range up to 4000 Hz [9]. Anyway, these experimental results are not entirely comparable with the effect that the absence of the TC simulated in our FEM would have, obviously opening the TC is a variation in the conditions of the AS, but the cavity remains producing reflecting waves, the air remains and presents some resistance and/or resonances to eardrum deformation. Two experimental paper established a relation between the middle ear cavities and a second resonance in gerbils [10, 11]. In addition, another paper [12] establishes a relation between the vibrations of gerbil TM and the opening of the middle ear cavity.
This paper is based on outer and middle ear numerical simulations developed by means of FEM. The maximum number of possible combinations of different outer and middle ear subsystems have been modeled and simulated: EEC, TM, OC, TC and simplified cochlea (SC). Four main different combinations have been modeled: EEC only attached to TM; EEC coupled to TM and TC; EEC coupled to TM, OC, and SC; and the full model: EEC coupled to TM, TC, OC and SC. Other three combinations have been simulated on the basis on the full model: The TC was opened to an air-filled domain with open boundary conditions that do neither reflect nor dampen the outgoing waves. These situations try to simulate experimental setups at the laboratory tests.
The model presents a complete fluid–structure interaction among EC, TM, TC and oval window. The TM modeling presents a crucial innovation respect to previous models; the elements used in the TM have an improved formulation, the formulation enhanced strain [13], which eliminates the problems of “shear locking” of the elements used in thin membranes. This fact together with a proper mesh convergence analysis provide sufficient and necessary guarantees of correct results in middle ear transfer functions. The results shown in this work are the EC and middle ear transfer functions as follows: EC, the ratio of pressure along the EC and TC to that of the EC entrance (pressure gain); middle ear: the ratio of umbo displacement to tympanic membrane pressure (UD/TMP); the ratio of Umbo velocity to tympanic membrane pressure (UV/TMP); the ratio of stapes displacement to tympanic membrane pressure (SD/TMP); and the ratio of stapes velocity to tympanic membrane pressure (SV/TMP).
Before continuing with the following section of the paper it must be stated what our numerical model does and what it does not. Which results are useful and which can only be considered qualitative.
Apart from geometry uncertainties dues to natural variability, the maximum level of inaccuracy is due to the difficulty to obtain the mechanical properties of some components (TM, tensors, joints…). In this paper, these values have been obtained from bibliography and has been assumed. There is no attempt to discuss its accuracy. In previous works [14], they have been object of a sensitivity analysis in order to establish its influence over the final results. Basically there are two different effect that we can group in two main features, those influencing the stiffness of the system (basically Young modulus) and those that alter the damping (viscoelasticity, damping, acoustic absorption). Even when they have been carefully chosen according to literature, as will be stated in the next section, their potential inaccuracy does not present a great influence and do not affect the main finding of this study.
Despite this, there are two key aspects in the numerical model that must be remarked due to its strong influence on the mechanisms studied. The first one is the proper fluid–structure interaction modeling. Even when there could be some delay effect (phase differences) due to some material viscoelasticity, it is negligible compared with the reflection effect of the cavities and the TM. So this coupled effect must be correctly simulated.
The second one it is related with the difficulty of modeling and meshing the membrane. A proper structure interaction with the fluid at both sides of the membrane requires the use of solid elements. It is a task that can introduce important errors in the model, depending on the mesh size and the finite element formulation.