Korean J Leg Med.  2023 Feb;47(1):1-7. 10.7580/kjlm.2023.47.1.1.

Dental Age Estimation Using the Demirjian Method: Statistical Analysis Using Neural Networks

• Roh BY
• Lee JS
• Lim SB
• Ryu HW
• Jeon SJ
• Lee JH
• Seo YS
• Ryu JW
• Ahn JM

• KMID: 2539924
• DOI: http://doi.org/10.7580/kjlm.2023.47.1.1

Abstract

In children and adolescents, dental age estimation is performed with the development of the teeth. Various statistical analysis methods have been used to determine the relationship between age and dental maturity and develop an accurate method of age calculation. This study attempted to apply a neural network model for the statistical analysis of dental age estimation in children and evaluated its applicability. This study used 1196 panoramic radiographs of patients aged 3–16 years, and 996 and 200 were randomly classified into training and test sets, respectively. The dental maturity of the mandibular left teeth was evaluated using Demirjian's method, the neural network model using the backpropagation algorithm was derived using training sets, and the errors were evaluated using 100 radiographs of each male and female as test sets. In addition, multiple linear regression analysis was conducted on the same training set, and the error was calculated by applying it to the test set and comparing it with the error of the neural network model. In the neural network model, the mean absolute error (MAE) and root mean squared error (RMSE) were 0.589 and 0.783 in male subjects and 0.529 and 0.760 in female subjects, respectively. In the multiple linear regression model, the MAE and RMSE were 0.600 and 0.748 in male subjects and 0.566 and 0.789 in female subjects, respectively. When applying the neural network model to the statistical analysis of the dental developmental stage, the results were as accurate as those of conventional statistical analysis methods. This study’s approach is expected to be useful for estimating the ages of children.

Keyword

Age determination by teeth; Panoramic radiography; Tooth development; Neural networks