Otoacoustic emissions are signals generated by the cochlea, either in the absence of acoustic stimulation (spontaneous emissions) or in response to acoustic stimulation (acoustically-evoked emissions) or electrical stimulation (electrically-evoked emissions). They are a by-product of the electromechanical action of the outer hair cells. Acoustically evoked otoacoustic emission testing gives an objective indication of whether the cochlear amplifier is functioning normally; it is particularly important for neonatal screening of hearing. The mechanisms generating the emissions are not fully understood, so that optimal stimulus paradigms are not yet available. This project is solving these problems.
It is known that in response to acoustic stimulation with two tones of frequencies, f1 and f2, emissions contain distortion components, called distortion product otoacoustic emissions (DPOAEs). The largest amplitude DPOAE is located at a frequency of 2f1-f2. This component is generated at two places in the cochlea: 1) The first component, which is called the primary-source component, is generated near the tonotopic place of the f2 tone by the nonlinear mechanoelectrical transducers in the stereocilia of the outer hair cells responding to the two stimulus tones. 2) The second component, called the secondary-source component, is generated at the place of the distortion product frequency, 2f1-f2, by a process called linear coherent reflection. The two components are transmitted back to the input of the cochlea. The two components can interfere - destructively or constructively - resulting in a noisy signal at the cochlea input and, therefore, in the middle ear and external ear canal (Turcanu et al., 2009; Dalhoff et al., 2013).
We are developing techniques to reduce the noise in the recorded signal. The techniques rely on extracting the primary-source component, so that there is little evidence of interference in the processed signal. The techniques use a continuous f1 tone, but a pulsed f2 tone. The first technique, we called onset-decomposition (Vetesnik et al., 2009), samples the response in its onset phase, at a time before the secondary-source component begins to interfere. For acoustic stimulation, we found that this technique allows auditory thresholds to be predicted with a standard deviation of only 4 dB, compared with 12 dB for conventional methods where both tones are continuous. However, this technique has two disadvantages: 1) the relative latency of the two source components should be established in advance, and 2) the gain in accuracy is achieved at the expense of measurement time. To avoid these two problems, we developed a second technique which decomposes the response into pulse basis functions (Zelle et al., 2013), thus allowing the entire signal trace to be used, rather than just a single sample instant per f2 pulse. This technique allows both source components to be extracted and yields auditory-threshold estimates which are as accurate as the onset-decomposition technique, but with at least 80% reduction in measurement time. We are now further developing these techniques.
Information from the experiments is being incorporated into models of sound processing in the cochlea, to understand function and develop new diagnostic tools.
A model of the organ of Corti is being developed together with Professor Charles Steele. The model yields mechanical impedance data consistent with measured data (Scherer and Gummer, 2004a). Conversely, the model explains the experimental observation that the impedance is purely viscoelastic, without an inertial component. The model is being incorporated into a complete model of cochlear function.
Models of the organ of Corti require accurate models of the cochlear components. To this end, together with Mario Fleischer and Rolf Schmidt we have developed a model of the compliance properties of the basilar membrane which is based on it's anatomical and material properties (Fleischer et al., 2010). The model predicts tonotopic properties of the basilar membrane; in particular, the observation from neural data that there are two exponential mapping constants: one for the low-frequency (< 1 kHz) apical region of the cochlea and a smaller one for the more basal region. Models based on anatomical and material properties of cellular structures in the cochlea are being developed.
To develop more powerful stimulus and analysis techniques for the application of otoacoustic emissions to clinical diagnosis (Preyer et al., 2001), in collaboration with Ales Vetesnik we are developing physically based models for the generation and propagation of otoacoustic emissions. We have shown theoretically that current experimental evidence supports the hypothesis that the emissions are transmitted back to the input of the cochlea by transverse pressure differences across the cochlear partition, rather than by longitudinal compression waves (Vetesnik and Gummer, 2012).
Active middle-ear implants (AMEI) are devices that, much like conventional hearing aids, sense an acoustic input signal and transmit it with suitable amplification to the middle ear to restore hearing. Contrary to their conventional counterparts, AMEI convert the input signal into mechanical vibrations which are coupled directly to the ossicular chain or to the round window. Thus, the actuator must be implanted into the middle ear, whereas signal acquisition and processing as well as the battery reside behind the auricle or partially in the mastoid. AMEI are especially important in the rehabilitation of patients with chronic ear-canal or middle-ear complications, when conventional hearing aids cannot achieve satisfactory results. Although AMEI generally achieve good results, several problems remain. We investigate possibilities for improving performance and reliability of AMEI.
1. Improvements to the Vibrant Soundbridge (VSB; Med-EL, Innsbruck). For both sites where the VSB is most often implanted - the long process of the incus and the round-window niche -, coupling efficiency as verified postoperatively is variable. In collaboration with Sebastin Schraven and Robert Mlynski, in human temporal-bone experiments we investigate different methods of improving the attachment of the VSB to the ossicular chain (Schraven et al., 2013).
2. Development of a round-window implant. Since several years, the VSB is implanted at the round-window niche, a technique that allows treatment of cases of combined middle- and inner-ear hearing loss. Because its diameter is too large to allow fitting into the round-window niche, the gap between the actuator and the round-window membrane must be filled with intervening material such as fascia. We have investigated the coupling efficiency of various methods of coupling mechanical vibrations to the round-window membrane (Schraven et al., 2011; Schraven et al., 2012). The optimal configuration was found for direct coupling. In collaboration with Fraunhofer IPA, Stuttgart, and NMI, Reutlingen, and industrial partners, we developed an actuator that fits into the round-window niche and allows direct contact to the round-window membrane (Goll et al., 2013). In addition, for that implant we developed a system for optical signal and energy transmission through the tympanic membrane (shown in the film).
3-D representation
of a hydrodynamic cochlea
model
Optical signal and energy transmission for a round-window implant