Dense representative tooth landmark/axis detection network on 3D model

Artificial intelligence (AI) technology is increasingly used for digital orthodontics, but one of the challenges to automatically and accurately detect tooth landmarks axes. While many attempts have been made streamline orthodontic treatments by automatic technologies, detection axes still left experienced dental practitioners performed manually. This partly because sophisticated geometric definitions them, due large variations among individual across different types tooth. As such, we propose a deep learning approach with labeled dataset professional dentists landmark/axis on model that are crucial treatments. Our method can extract not only in form points (e.g. cusps), also measure angulation inclination. The proposed network takes as input 3D predicts various Specifically, encode dense fields defined surface model. design choice set added components make more suitable extracting sparse from given Extensive evaluation was conducted models prepared dentists. Results show our produce high accuracy. examined justified via comparison state-of-the-art methods well ablation studies.