High-resolution in vivo imaging of inner ear using photon-counting-detector computed tomography and artificial intelligence

PROJECT SUMMARY/ABSTRACT Disorders of the inner ear manifest as an array of audiovestibular diseases and constitute a significant public health burden, affecting millions of individuals and costing billions of dollars each year. As the population ages worldwide, the disease prevalence is intensifying, necessitating immediate action to address this growing public health issue. CT imaging of the inner ear is a central component of the diagnostic evaluation of patients presenting with vestibular symptoms, hearing loss, autophony, and chronic otitis media. As a part of this evaluation, accurate detection and characterization of potential labyrinthine dehiscence is imperative to definitively rule in or rule out disease, inform patient counseling, guide shared clinical decision-making, quantify the risk and potential outcome associated with surgical intervention, and further understand the natural history of disease. Notably, current in vivo imaging techniques are unable to consistently decipher thin versus truly dehiscent labyrinthine bone. Further, current CT imaging is not able to fully characterize dehiscence morphology (shape, size, and surface area). As such, high-quality, high-resolution, artifact-free in vivo imaging is critically needed for the reliable diagnosis and management of inner ear disorders such as superior semicircular canal dehiscence syndrome and cholesteatomatous labyrinthine fistula. To address this unmet clinical need, we propose to develop high-fidelity clinical imaging techniques for the inner ear using photon counting detector computed tomography (PCD-CT) and advanced artificial intelligence (AI) algorithms. Our team has been at the forefront of PCD-CT development and its clinical translation, and has an established track record of strong multidisciplinary collaborative innovation in this space. Building upon our successful prior work, we will harness the power of AI – in synergy with PCD-CT – to reduce image noise, improve spatial resolution, and reduce radiation dose for CT imaging of the inner ear. We will develop and validate these techniques with cadavers, and in vivo patient CT exams. The innovation of our proposal lies in the synergy of PCD-CT and advanced AI algorithms for spatial resolution improvement, noise and dose reduction, and a cadaver library of normal and labyrinthine thinning and dehiscence specimens that will enable iterative protocol optimization of ultra-high-spatial-resolution inner ear imaging. Without the proposed techniques, PCD-CT, while impressive, remains limited by noise, image artifacts, and radiation dose. Further, under controlled conditions using cadaveric specimens, definitive evidence of the impact of the developed techniques will be obtained. Successful completion of this proposal will have a significant impact on patients with inner ear diseases, allowing high-fidelity imaging with improved diagnostic accuracy and individualized patient management. In addition, the technical innovations for noise reduction, super-resolution, and dose reduction developed in the proposal will benefit inner ear imaging tasks beyond dehiscence, including direct visualization of the interscalar cochlear partition, and can be readily applied to the middle ear.