Application of Artificial Intelligence in Automatic Cephalometric Landmark Annotations—A Comprehensive Review

Cephalometric analysis, a frequently used orthodontic clinical routine, is becoming a prime research area for artificial intelligence (AI)-based solutions in dentistry. The need for AI-based tools in dentistry increased efforts to make cephalometric measurements quick and accurate. This review presents a qualitative and comprehensive study of AI-based solutions and existing tools for cephalometric diagnosis. Research publications from the past ten years are collected from electronic databases like Scopus, Google Scholar, and PubMed for review. It is inferred from the studies that large numbers of research groups contribute to the area. Recent studies using deep learning approaches reported increased landmark localization accuracy from 80 to 96%. Non-availability of a large public dataset is a limitation and hinders the relative comparison of the existing method’s performance. The convolutional neural network is the best architecture with a high success detection rate (SDR) and significantly less mean radial error (MRE) in the 2 mm precision range. The review aids in grasping AI’s significance in cephalometric analysis and pinpoints areas for future research.