Design of Sensors and Actuators for Totally Implantable Auditory Prostheses
Panagiota Kitsopoulos · Deep Blue (University of Michigan) · 2025
Hearing loss is a debilitating condition that affects over 5% of the world’s population. Hearing aids and cochlear implants help patients treat their hearing loss, but have limitations impacting their use rates. Totally implantable auditory prostheses would expand the range of activities a prosthesis user could engage in and enable 24/7 use. However, the lack of a completely implantable microphone that is robust, lightweight, and low-noise prevents the widespread adoption of completely implantable auditory prostheses. Current implantable sensors struggle to meet or exceed the performance necessary for this application. Moreover, three active middle ear implants have been developed to address some of the barriers associated with conventional hearing aids. However, all devices rely on transducers mechanically attached to a middle ear structure to deliver amplified stimuli, with no other alternative options. In this work, we aim to provide solutions that help bridge these gaps. First, we developed a discretized and exhaustive design optimization approach to identify multi-bandwidth transducers that meet the 20-phon noise floor over 100 Hz – 8 kHz, and can replace the external microphones in implantable prostheses. Our design procedure is based on an experimentally validated analytical model that simulates the response of miniature piezoelectric microelectromechanical systems (MEMS) accelerometers. A four-bandwidth accelerometer with constrained proof mass thicknesses is selected as the design that best balances area minimization with ease of manufacturability. The estimated MEMS die dimensions are 825 μm x 575 μm, which is a 23% MEMS die area reduction compared to the previously published dual-bandwidth sensor. Second, we investigated the effect of the piezoelectric thickness, loss tangent, and piezoelectric coefficient on the sensor area of three canonical sensor designs: a sensor under uniform pressure, a sensor with a prescribed displacement at the free end, and an accelerometer with a proof mass. All three sensor designs have been or are currently used in research for totally implantable auditory prostheses. This study identified the direct relationships between the sensor areas and the piezoelectric thickness. These relationships can be used to predict the sensor area given a particular piezoelectric thickness, noise floor, and operational bandwidth. Specifically, assuming the material properties vary with thickness, the sensor area for a sensor under uniform pressure remains approximately constant regardless of the thickness. In contrast, the sensor areas for a sensor with a prescribed displacement and for an accelerometer decrease with an increasing thickness. However, these results depend on the material properties, and, for the accelerometer design, also on the operational bandwidth. Lastly, we carried out a new analysis where we investigated the feasibility of using an implanted middle ear speaker to amplify the motion of the stapes. We identified that an implanted middle ear speaker with a negative gain and small time delays can be used to provide amplification at the stapes. Specifically, at high frequencies (>1 kHz), with a gain of -1, and no feedback or time delays, the amplification can be up to six times compared to the motion of the stapes with no middle ear speaker. Even greater amplification can be achieved with more negative gains. However, accounting for a time delay of 10 ms, we identified that for gains less than -4.5, the system can become unstable. In addition, high time delays introduce oscillations in the displacement of the stapes, but low time delays can significantly reduce this effect.