Should we consider the sex when estimating bone age from hand bone biometrics?

Abstract

Recently, it was proposed to estimate age from the biometric information of hand bones. We observed that these estimations became less precise as children get older, especially from the age of 13-15years. This study aimed to evaluate the influence of considering sex for age estimation based on hand bones biometrics. The study sample consisted of metacarpals and proximal phalanges measurements collected on 1003 medical images performed at Nancy and Marseille Hospitals of individuals aged under 21years. This sample was divided into two subgroups delineated by the age of 13, as it is a relevant legal threshold for most European countries. First, the influence of sex on the hand bones biometrics and on the estimated age was evaluated. Then, based on these results, new sex-specific age estimation formulas were constructed from linear models and their precision was assessed using residual analysis, in comparison with previous global formulas. An influence of sex was only highlighted from the age of 13 and for the total study sample. Thus, new sex-specific age estimation formulas were built for the [1-21] global sample and the [13-21] subsample. Even though the differences with the previous formulas were minor, age was more accurately estimated when sex was considered. Considering sex in age estimation is relevant when relying on hand bone biometrics. A new tool was proposed to select the most appropriate age estimation formula, based on the discriminant analysis result and the a priori knowledge of the sex.