TL;DR
Researchers have developed the "DrumMic," a piezoelectric membrane microphone designed to be implanted in the middle ear to sense vibrations at the tip of the malleus (the umbo). Tested in human cadaveric temporal bones, the device demonstrates robust, linear performance across speech frequencies, bringing us a significant step closer to totally implantable cochlear implants that preserve the natural acoustic benefits of the outer ear.
What Problem Does This Paper Address?
While cochlear implants (CIs) have revolutionized hearing for over a million people, they still rely on external hardware—specifically microphones worn behind the ear. This setup has several drawbacks: it’s aesthetically conspicuous, susceptible to wind noise, unusable in water or during sleep, and, perhaps most importantly, it bypasses the natural "pinna effect." The outer ear (pinna) and ear canal naturally filter and amplify sound, providing crucial cues for localization and speech clarity in noise.
The "holy grail" of CI research is a totally implantable system. However, the bottleneck has always been the microphone. Subcutaneous microphones (placed under the skin) suffer from poor signal-to-noise ratios because the skin attenuates high-frequency sounds while picking up internal body noise. This paper explores a middle-ear approach, aiming to create a sensor that is sensitive, low-noise, and easy to integrate with existing surgical procedures.
Methods
The authors introduced a "DrumMic" concept based on a piezoelectric polymer called polyvinylidene fluoride (PVDF).
- Design: The microphone consists of a thin PVDF membrane stretched over a 3mm brass ring. One side of the membrane presses against the umbo (the tip of the malleus bone attached to the eardrum), while the other side sits on the cochlear promontory.
- Mechanism: As sound vibrates the eardrum and the ossicular chain, the umbo moves, stretching the PVDF membrane. This mechanical strain generates an electrical charge, which is then processed by a custom low-noise charge amplifier.
- Experimental Setup: The team tested five different DrumMics across four human cadaveric temporal bones. They used Laser Doppler Vibrometry (LDV) to measure the actual velocity of the umbo and compared it to the electrical output of the DrumMic.
- Adjustability: To ensure good mechanical contact, they used tiny plastic shims (0.65 mm thick) to "tension" the microphone between the bone and the umbo.
Datasets
This study did not use a traditional computational dataset. Instead, it generated primary experimental data from:
- Five custom-fabricated DrumMics.
- Four human cadaveric temporal bones (harvested from four male donors, mean age 71.5, with no history of ear pathology).
- Comparative benchmarks: Performance was compared against the Knowles EK-23103, a standard high-end microphone used in commercial hearing aids.
Key Insights and Ideas
- The Umbo as a Signal Source: The researchers chose the umbo because it exhibits the largest, most predictable one-dimensional motion in the middle ear, making it an ideal "mechanical" output for sound.
- The "Canceling" Effect: A fascinating finding was that the DrumMic's sensitivity to ear canal pressure remained consistent even when the anatomy varied. While a tighter fit (more shims) increases the tension and sensitivity of the PVDF film, it also stiffens the ear's mechanics, reducing the umbo's motion. These two factors effectively cancel each other out, resulting in a device that is remarkably robust to surgical placement variations.
- Leveraging Natural Anatomy: By placing the mic behind the eardrum, the system benefits from the natural 10–20 dB pressure gain provided by the ear canal and pinna, which helps offset the intrinsic noise of the sensor.
Key Results
- Linearity: The DrumMic behaved linearly across a massive dynamic range, from 46 dB SPL to at least 100 dB SPL.
- Bandwidth: The device maintained a strong signal-to-noise ratio up to 8 kHz, covering the essential frequencies for human speech.
- Sensitivity: The output was tightly clustered around 25 fC/Pa (femtocoulombs per Pascal) across different ears and devices.
- Noise Floor: The measured Equivalent Input Noise (EIN) was 54 dB SPL. However, when accounting for the natural gain of the outer ear, the effective EIN was estimated at 46 dB SPL. While this is higher than the ~34 dB SPL of top-tier external hearing aid mics, it is a significant improvement over many previous implantable concepts.
Limitations and Future Work
- Environmental Protection: The middle ear is a moist environment. The current prototype was only lightly coated in silicone; a final version will require sophisticated hermetic sealing and medical-grade biocompatibility.
- Anatomical Interference: The base of the current design sits near Jacobson’s nerve. Future iterations must ensure the device does not cause trauma to this or other middle-ear structures.
- Noise Reduction: While 46 dB SPL is promising, further optimization of the charge amplifier and electrical shielding is needed to match the performance of external microphones.
- Long-term Stability: The study used cadaveric bones; future work must address how the mechanical coupling holds up over years of vibration and pressure changes in a living patient.
Summary updated