Current Applications and Emerging Horizons of Artificial Intelligence in Temporal Bone Computed Tomography

The temporal bone has intricate anatomy. CT examinations can clearly depict fine anatomical structures and subtle lesions, providing indispensable imaging evidence for diagnosing otologic diseases and planning surgical procedures. However, image interpretation has several inherent limitations, including subjectivity, substantial interobserver variability, and heavy reliance on physician experience, all of which can compromise clinical decision-making and lead to inconsistent patient management. In recent years, AI-driven approaches have made significant progress in this field, particularly in optimizing image quality, identifying key anatomical structures, and assisting in the diagnosis of common otologic diseases. These advances span multiple tasks and have gradually narrowed the performance gap between automated tools and expert readers. Despite encouraging results in research settings, AI tools continue to encounter multiple obstacles in the transition from algorithmic validation to routine clinical use. These challenges include insufficient and imbalanced training data, limited generalizability across diverse populations and imaging protocols, and a persistent lack of clinician trust arising from the "black-box" nature of these models. Moreover, the absence of standardized benchmarks and prospective clinical trials further impedes the deployment of these technologies in routine otologic practice. This paper comprehensively reviews current AI applications in temporal bone CT imaging and discusses related emerging technologies, aiming to provide new perspectives and methods for clinical practice.