Dental age estimation of Malay children and adolescents: Chaillet and Demirjian's data improved using artificial multilayer perceptron neural network

• Main Authors: Bunyarit, Safar Sumit, Nambiar, Phrabhakaran, Naidu, Murali K., Ying, Rozaida Poh Yuen, Asif, Muhammad Khan
• Format: Article
• Published: Elsevier 2021
• Subjects: RK Dentistry; Oral surger
• Online Access: http://eprints.um.edu.my/34920/

Purpose: The aim of this study was to assess the suitability of the 8-tooth method on the Malaysian Malays aged 5.00-17.99 years originally introduced by Chaillet and Demirjian and to develop a more reliable teeth maturity scores for age estimation based on artificial neural networks (ANN). A total of 1569 dental panoramic tomographs of Malaysian Malays aged 5.00-17.99 years were assessed. Tooth development was initially recorded using a score chart adapted from Demirjian's eight developmental stages. Maturity scores were converted to dental age (DA). The mean and standard deviation of chronological age (CA), DA and the difference between CA and DA were calculated. Artificial neural networks was then employed to develop new dental maturity scores (NDA) and its reliability was assessed. Results: Initial comparison of known CA with estimated DA derived from Chaillet and Demirjian's dental maturity scores showed that DA was consistently underestimated by 2.09 +/- 0.90 years for Malay boys and 2.79 +/- 0.99 years for Malay girls for all age groups (paired t-test, p < 0.05). In order to improve the age estimation, an adaptation using artificial neural networks was employed to develop new dental maturity scores (NDA) that was more suitable for the Malay subjects. The results showed more accuracy in estimating dental age (0.035 +/- 0.84 years for boys and 0.048 +/- 0.928 years for girls, paired t-test, p > 0.05). Conclusion: In conclusion, the new dental maturity scores were highly accurate in estimating the age of Malaysian Malay children and adolescents. (C) 2021 Japanese Society of Pediatric Dentistry. Published by Elsevier Ltd. All rights reserved.