Högskolan i Skövde

Background: Bone age assessment (BAA) is used in numerous pediatric clinical settings, as well as in legal settings when entities need an estimate of chronological age (CA) when valid documents are lacking. The latter case presents itself as critical since the law is harsher for adults and granted rights along with imputability changes drastically if the individual is a minor. Traditional BAA methods suffer from drawbacks such as exposure of minors to radiation, do not consider factors that might affect the bone age and they mostly focus on a single region. Given the critical scenarios in which BAA can affect the lives of young individuals it is important to focus on the drawbacks of the traditional methods and investigate the potential of estimating CA through BAA.

Objective: This paper aims to investigate CA estimation through BAA in young individuals of 14 to 21 years with machine learning methods, addressing the drawbacks in the research using magnetic resonance imaging (MRI), assessment of multiple ROIs and other factors that may affect the bone age.

Methods: MRI examinations of the radius, distal tibia, proximal tibia, distal femur and calcaneus were carried out on 465 males and 473 females subjects (14-21 years). Measures of weight and height were taken from the subjects and a questionnaire was given for additional information (self-assessed Tanner Scale, physical activity level, parents' origin, type of residence during upbringing). Two pediatric radiologists assessed, independently, the MRI images as to their stage of bone development (blinded to age, gender and each other). All the gathered information was used in training machine learning models for chronological age estimation and minor versus adults classification (threshold of 18 years). Different machine learning methods were investigated.

Results: The minor versus adults classification produced accuracies of 90% and 84%, for male and female subjects, respectively, with high recalls for the classification of minors. The chronological age estimation for the eight age groups (14-21 years) achieved mean absolute errors of 0.95 years and 1.24 years for male and female subjects, respectively. However, for the latter lower error occurred only for the ages of 14 and 15.

Conclusions: This paper proposed to investigate the CA estimation through BAA using machine learning methods in two ways: minor versus adults classification and CA estimation in eight age groups (14-21 years), while addressing the drawbacks in the research on BAA. The first achieved good results, however, for the second case BAA showed not precise enough for the classification.