Background: The BoneXpert method is an automated determination of bone age, which employs deformable models of each bone to locate the bones, and extracts the component of the bone appearance related to maturity in a holistic, statistical manner. The BoneXpert has been validated on normal children and children with diagnoses typical of pediatric endocrinology. Multiple clinical studies suggested that BoneXpert has adequate accuracy, precision, and efficiency to be clinically useful.
Objective: To investigate accuracy, precision and bias of conventional methods of bone age evaluation compared with BoneXpert.
Design: Comparison of bone age assessed using the method of Greulich and Pyle, Tanner-White-house (TW-RUS) or BoneXpert. 79 radiographs of the left hand were obtained from children with various diagnoses between the ages of 4.8 to 16.6 yr. The films were rated using the GP atlas method by two different raters, TW-RUS method by two different raters and BoneXpert. Results are evaluated using Rank and Intraclass correlation, and Bland-Altman analysis.
Results: Human GP (HGP) versus BoneXpert GP (BGP) score shows bias towards chronological age for HGP ratings, HGP scores being on average 0.3 years higher than BGP scores. We noticed similar results when comparing HGP with TW-RUS scores. There was no evidence of bias in TW-RUS scores towards chronological age. TW-RUS scores were on average 0.51 years higher than BGP scores. Rank and intraclass correlation stats showed high correlation between all methods. TW RUS showed higher correlation with BGP than HGP with BGP.
Conclusion: The results of our study suggest that BoneXpert has adequate accuracy, precision and efficiency. The automated method provides a reliable and efficient standard for bone age determination. Introduction of BoneXpert into clinical practice provides precise standardized bone age determination, vital for long term outcome and comparator studies of growth and growth interventions.
27 - 29 Sep 2018
European Society for Paediatric Endocrinology