Research Progress and Application of Artificial Intelligence in Cephalometric Analysis

Cephalometric analysis is a vital diagnostic tool in orthodontics and craniofacial surgery. It provides precise information on the relationship between the skull and teeth to guide treatment planning. Unlike traditional manual annotation methods, artificial intelligence (AI) approaches not only enable more efficient annotation but also mitigate result variability caused by differences in operator skill and experience. AI, particularly deep learning, excels at rapidly processing image while achieving accuracy comparable to that of experts. Recent studies confirm the effectiveness of convolutional neural networks and related architectures in automated landmark detection, structural segmentation, and intelligent measurement-based diagnostic support. When trained on large-scale annotated datasets, these models extract stable anatomical and pathological features. This enhances result reproducibility, reduces analysis time, and improves clinical efficiency. However, there are also some challenges persist, including inconsistent annotation protocols, limited model generalization across populations and imaging devices, and a lack of large-scale external validation. This review aims to summarize AI applications in cephalometric analysis, evaluate existing limitations, and explore future directions for establishing standardized, clinically reliable applications.