Abstract
Humans are highly dependent on the ability to process audio in order to interact through conversation and navigate from sound. For this, the shape of the ear acts as a mechanical audio filter. The anatomy of the outer human ear canal to approximately 15-20 mm beyond the Tragus is well described because of its importance for customized hearing aid production. This is however not the case for the part of the ear canal that is embedded in the skull, until the tympanic membrane. Due to the sensitivity of the outer ear, this part, referred to as the bony part, has only been described in a few population studies and only ex-vivo. We present a study of the entire ear canal including the bony part and the tympanic membrane. We form an average ear canal from a number of MRI scans using standard image registration methods. We show that the obtained representation is realistic in the sense that it has acoustical properties almost identical to a real ear.
Introduction
Designing devices that sit in the ear, such as hearing aids or high-end monitors, requires an intimate understanding of ear canal geometry. However, a significant "blind spot" exists in current anatomical data. Traditional physical impressions using silicone can only safely reach about 15–20 mm into the canal. Beyond this point lies the "bony part"—a region so sensitive that physical contact can cause intense pain or trigger a fainting response via the auriculotemporal nerve.
Consequently, much of our knowledge regarding the deep ear canal and the tympanic membrane (eardrum) has historically relied on cadaver studies. This research addresses the lack of accurate, in vivo data for the entire ear canal and investigates whether a mathematical "average" shape can capture the complex acoustical properties of a diverse population.
Methodology
The researchers employed a multi-disciplinary approach combining medical imaging, geometry processing, and acoustic simulation.
Data Acquisition via Contrast-Enhanced MRI
To visualize the "empty" space of the ear canal, the canals of 44 volunteers were filled with rapeseed oil. On a T2-weighted MRI, the oil appears bright against the dark surrounding tissue, providing a high-resolution map of the cavity.
Shape Analysis and Registration
After manually segmenting the MRI slices to create 3D surfaces, the team utilized Iterative Closest Point (ICP) and non-rigid registration (B-spline Free-Form Deformation). This process mathematically warped every individual ear shape into a common reference frame to calculate a "Frechet mean"—the most representative average shape.
Acoustical Validation
To validate the model, the team used Finite Element Modeling (FEM) to simulate sound wave propagation within the digital average ear. These simulations were compared to physical wide-band impedance measurements (35 Hz to 25 kHz) taken from the study participants using custom-molded ear tips.
Key Insights
- Acoustic Realism of the Average: Unlike many biological models where averaging results in a non-functional "blurry" shape, the simulated impedance of this average geometry was almost identical to the median measured impedance of the living subjects. This confirms the model as a valid proxy for the population.
- Low Anatomical Variance: Principal Component Analysis (PCA) revealed that the human ear canal is surprisingly consistent, with just six components explaining more than 80% of the shape variance in the dataset.
- Anatomical Fidelity: The model successfully recovered features often lost in traditional impressions, including the first and second bends of the canal, the specific slant of the tympanic membrane, and the attachment point of the Malleus (the "hammer" bone of the middle ear).
Results and Discussion
The resulting 3D model provides a clear, parameterized representation of the human ear canal from the Tragus to the eardrum. The magnitude of the impedance output from the simulation closely tracked the human median impedance across the full audio band, particularly in the critical 1 kHz to 10 kHz range where ear resonance is most prominent. By comparing the average shape to the "closest individual ear" in the dataset, the researchers demonstrated that the average model is a robust parametrization for simulating audio device performance.
Conclusion and Future Work
While the study provides a high-fidelity model, it was limited to 44 Caucasian adults. Future research should expand the demographic scope to include different ethnicities, ages, and genders to ensure global representativeness. Additionally, incorporating the dynamic nature of the ear canal—which changes shape during activities like talking or chewing—remains a key frontier for precision audio engineering and next-generation hearing aid design.
Summary updated