Radiographic dental age estimation of Brazilian children and adolescents using a national reference sample

Abstract

Several radiographic methods have been proposed for dental age estimation, among which the Willems model and its Brazilian adaptation by Franco et al. (2024) have shown promising results. However, the Brazilian model has not yet been tested in Northeastern populations. This study aimed to validate Franco's model in a sample of children and adolescents from Alagoas, Northeastern Brazil. A retrospective, cross-sectional study was performed on 500 panoramic radiographs (250 males, 250 females) aged 6–15.99 years, distributed evenly across one-year intervals. Chronological age was calculated from documentary records and compared with dental age estimated using Demirjian's staging system and Franco's Brazilian model. Mean error (ME), mean absolute error (MAE), and root mean squared error (RMSE) were calculated and complemented by Bland – Altman analysis. The mean chronological age was 11.0 ± 2.9 years, while the mean estimated age was 10.8 ± 2.5 years. The overall ME was 0.18 years (0.06 males, 0.29 females). The MAE was 0.70 years (0.67 males, 0.73 females), with the best performance between 7 and 10.99 years (< 0.5 years). RMSE values increased in older groups, reaching > 1.0 year after age 13.99. Bland–Altman plots showed a transition from overestimation in younger ages to underestimation in older ages, earlier in females. The Brazilian model was adequate for the studied population, particularly for individuals aged 7–10.99 years, with overall differences below one year.