Experimental Framework for the Setup and Validation of Individualized Bone Conduction Hearing Computational Models

TL;DR

This paper introduces a framework for creating and validating personalized computational models of bone conduction hearing, using multi-level anatomical data and experimental validation.

Contribution

The study presents a hierarchical validation framework for subject-specific bone conduction models, enabling scalable and experimentally supported simulations.

Findings

Photon-counting CT outperformed energy-integrating CT in structural resolution, especially at the full-head level.

Vibrational measurements showed consistent resonance patterns across different anatomical levels.

Hydration state and thickness reduction significantly influenced vibrational behavior.

Abstract

In bone conduction (BC) hearing, sound is transmitted directly to the cochlea via skull vibrations, bypassing the outer and middle ear. This provides a therapeutic option for patients with conductive or mixed hearing loss and single-sided deafness. Although finite-element models have advanced understanding of the mechanisms underlying BC, progress toward personalized treatment strategies remains limited by a lack of standardized, experimentally validated, subject-specific models. This study proposes a hierarchical validation framework to support the development and validation of individualized computational models of the human head under BC stimulation. The framework spans four anatomical levels: system, subsystems, structures, and tissues. This approach enables systematic acquisition of data from intact cadaver heads down to isolated material domains. To demonstrate the applications of the framework, an experimental study was conducted on a single cadaver head, targeting three levels: the intact head (system), extracted bone pieces (structures), and isolated cortical layers (tissues). Subsystems were not addressed. High-resolution photon-counting computed tomography (CT) and energy-integrating cone-beam CT were used to acquire anatomical data. One-dimensional laser Doppler vibrometry was used to capture vibrational responses of bone pieces and cortical layers under wet and dry conditions. Representative results were analyzed to assess the impact of preparation state on resonance behavior. Comparative analysis showed that photon-counting CT provided superior structural resolution compared with energy-integrating cone-beam CT, particularly at the full-head (system) level. Vibrational measurements at the structure and tissue levels from the same anatomical region revealed broadly consistent resonance vibration patterns, enabling comparison of resonance frequencies. The influence of hydration state and thickness reduction on vibrational behavior was highlighted. The proposed framework provides a scalable methodology for validation of subject-specific BC models with the potential for more accurate BC simulations based on the hypothesis of functional variability rooted in anatomical variability. Obvious use cases would include the development of improved hearing aid designs and personalized treatments. In parallel, a successful correlation of anatomical and functional variability can serve as inspiration for design principles of metamaterials.

Figures