Background: Skeletal maturation is the most reliable indicator of biological age in children and adolescents. The evaluation of hand and wrist X-Ray according to Tanner-Whitehouse (TW3) or Greulich-Pyle (GP) are the most commonly used methods for biological age assessment. Automated bone age assessment has recently become increasingly popular, however a large independent study comparing automated and manual evaluation of bone age is still missing. The aim of this study was to assess the differences between automated and manual evaluation of bone age using TW3 and GP method.
Methods: In this cross-sectional study we evaluated bone age scans using TW3 and GP methods in 1285 children and adolescents (659 boys, range 5.015.9 years, median 10.3, IQR 4.9 years) with various endocrine conditions in parallel manually and using BoneXpert software (Visiana, Holte, Denmark). Root mean square errors (RMSE) were calculated for the whole group and for sex-specific one-year age categories (girls between 5 and 15 years, boys between 5 and 16 years, over 50 children in each category).
Results: In total RMSE were 0.61 years and 0.58 years in boys and 0.79 years and 0.60 years in girls, respectively for TW3 and GP. Sex- and age-specific analysis showed the greatest differences between manual and automated TW3 evaluation in girls between 67, 1213 and 1314 years with RMSE 0.90, 0.90 and 1.05 years, respectively. Manual and automated evaluation differed by more than 1 year in 9.7% and 7.0% boys and 18.2% and 8.6% girls, respectively for TW3 and GP.
Conclusion: Automated bone age assessment provides sufficient agreement with manual evaluation in most scans of children with common endocrine disorders. Bone age assessment provided by BoneXpert tends to be underestimated, especially in girls during puberty using TW3 method. Further analysis is required to indentify the source of these differences.
27 - 29 Sep 2018
European Society for Paediatric Endocrinology