A Deep Learning-Based System for the Assessment of Dental Caries Using Colour Dental Photographs Maryam MEHDIZADEH a , Mohamed ESTAI a,b , Janardhan VIGNARAJAN a , Jilen PATEL c,d , Joanna GRANICH e , Michael ZANIOVICH f , Estie KRUGER b , John WINTERS d , Marc TENNANT b , and Sajib SAHA a a The Australian e-Health Research Centre, CSIRO, Kensington, Australia b School of Human Sciences, The University of Western Australia, Crawley, Australia c UWA Dental school, The University of Western Australia, Crawley, Australia d Department of Pediatric Dentistry, Perth Children Hospital, Nedlands, Australia e Telethon Kids Institute, The University of Western Australia, Crawley, Australia f Aria Dental, Perth, Australia ORCiD ID: Maryam Mehdizadeh https://orcid.org/0000-0003-2164-828X Abstract. D 1 ental caries remains the most common chronic disease in childhood, affecting almost half of all children globally. Dental care and examination of children living in remote and rural areas is an ongoing challenge that has been compounded by COVID. The development of a validated system with the capacity to screen large numbers of children with some degree of automation has the potential to facilitate remote dental screening at low costs. In this study, we aim to develop and validate a deep learning system for the assessment of dental caries using color dental photos. Three state-of-the-art deep learning networks namely VGG16, ResNet-50 and Inception-v3 were adopted in the context. A total of 1020 child dental photos were used to train and validate the system. We achieved an accuracy of 79% with precision and recall respectively 95% and 75% in classifying `caries' versus `sound' with inception-v3. Keywords. Oral health, dental caries, remote health 1. Introduction Dental caries, commonly referred to as tooth decay, remains the leading childhood chronic disease, affecting almost half of children's population [1]. The temporary shutdown of dental services due to COVID-19 is likely to have worsened the already significant dental disease burden and further overburdens the health systems. To assist in visual examination of dental caries, recent research has used smart phone photos as input data to deep learning models [2]. If deep learning technology is combined with a smartphone camera, this has the potential to play a vital role in expediting remote dental screening and early classification of tooth decay, even during the times of crisis- related dental services shutdown. 1 Corresponding Author: Maryam Mehdizadeh, maryam.mehdizadeh@csiro.au MEDINFO 2023 — The Future Is Accessible J. Bichel-Findlay et al. (Eds.) © 2024 International Medical Informatics Association (IMIA) and IOS Press. This article is published online with Open Access by IOS Press and distributed under the terms of the Creative Commons Attribution Non-Commercial License 4.0 (CC BY-NC 4.0). doi:10.3233/SHTI231097 911