Frontiers in endocrinology · 2026
digiBONE: an automated tool for segmental Greulich-Pyle bone age assessment of Indian children and adolescents.
СтатьяНазвание и аннотация: Перевод в очереди
Обработка статьи · этапы, ошибки и управление
Загрузка состояния…
Ошибки и ограничения
История · было → стало
Состояние обновляется каждые 5 секунд, пока блок открыт. Процент показывается только там, где известен объём работы. Пауза останавливает продолжение этапа; текущий запрос к источнику может завершиться.
Chakladar Shreya, Oza Chirantap, Mondkar Shruti, Aeppli Tim R J, Sävendahl Lars, Khadilkar Anuradha, Khadilkar Vaman, Goel Pranay
Аннотация
INTRODUCTION: Accurate bone age assessment (BAA) is essential for diagnosing and managing pediatric endocrine and growth disorders, as it reflects biological maturity beyond chronological age. The widely used Greulich-Pyle (GP) method estimates bone age by visually comparing full-hand radiographs with standardised reference images. Although widely used, this technique-and most automated systems based on it-assumes uniform skeletal maturation across the hand. In practice, however, skeletal maturation progresses at different rates in the anatomical segments of the hand under varied hormonal influences. This segmental variability may contribute to inter-observer inconsistency and diagnostic uncertainty. We developed digiBONE, a deep learning framework that models segment-specific skeletal maturation.
METHODS: Hand radiographs were segmented into anatomically coherent regions-short bones, carpals, and wrist-representing synchronously maturing bones. Separate convolutional neural networks (CNNs) were trained for each segment and for the full hand. Segmental predictions were combined with the full-hand model to generate a composite bone age prediction.
RESULTS: Integration of segmental maturity improved performance over full-hand-only models, achieving mean absolute differences (MAD) of 4.75 months for boys and 4.93 months for girls. In addition to improved accuracy, it revealed asynchrony between hand regions providing complementary maturity information beyond global estimates.
DISCUSSION: The full-hand model captures how a child's overall maturity aligns with population norms, while, the segmental models explain how individuals of the same GP age differ biologically, improving interpretability and personalisation. digiBONE demonstrates that integrating biologically relevant segmental information into deep learning pipelines offers a scalable, automated bone age assessment solution applicable to Indian populations.
METHODS: Hand radiographs were segmented into anatomically coherent regions-short bones, carpals, and wrist-representing synchronously maturing bones. Separate convolutional neural networks (CNNs) were trained for each segment and for the full hand. Segmental predictions were combined with the full-hand model to generate a composite bone age prediction.
RESULTS: Integration of segmental maturity improved performance over full-hand-only models, achieving mean absolute differences (MAD) of 4.75 months for boys and 4.93 months for girls. In addition to improved accuracy, it revealed asynchrony between hand regions providing complementary maturity information beyond global estimates.
DISCUSSION: The full-hand model captures how a child's overall maturity aligns with population norms, while, the segmental models explain how individuals of the same GP age differ biologically, improving interpretability and personalisation. digiBONE demonstrates that integrating biologically relevant segmental information into deep learning pipelines offers a scalable, automated bone age assessment solution applicable to Indian populations.