Japan Association for Medical Informatics

[3-H-3-04] Development of an analysis application to distinguish the state of the tooth from intraoral photographs: First report.

*Atsuhiro Kinoshita1, Hideki Matsui2, Satoru Takigami2, Tetsushi Kawabe2, Tomoya Asanuma3, Yukiko Uno3, Chiaki Mikawa3, Hiroto Oka3, Atsushi Goto3 (1. Tokyo Medical and Dental University, 2. CANON MEDICAL SYSTEMS CORPORATION, 3. CANON INC.)

deep learning, oral examination standard code, application development

Dentists must record intraoral findings on dental treatment records at patients' first visit. It takes substantial time to record the presence and status of all 32 teeth. On the other hand, during the Great East Japan Earthquake of 2011, the usefulness of the identification with dental treatment records and the importance of recording tooth numbers and status electronically in dental practices was confirmed. Furthermore, the "Japan Dental Association Oral Examination Standard Code" was adopted as a standard code of the Ministry of Health, Labour and Welfare in March 2021, and "Basic Law on the Acceleration of Probes into the Cause of Death and Identification" stated that the State and any local public entity should promote the standardization of information on dental practices and complete the databases. Therefore, the importance of electronic dental treatment records is increasing more and more. Many studies and approaches using deep learning with dental X-ray images were performed to improve efficiency in dental clinics and digitalize intraoral information. We are developing an application software identifying tooth numbers and status with deep learning from intra-oral photographs taken in dental clinics and exporting the results as the "Japan Dental Association Oral Examination Standard Code" to electronic dental health record systems. In this presentation, we report the ongoing progress of our study.