Objective:

To develop a deep learning (DL)-based algorithm for accurate and reproducible measurement of the femoral head ossification center (FHOC) size and establish a radiographic growth chart for healthy children, enhancing clinical assessments.

Key Findings:

• The DL algorithm demonstrated high accuracy in segmenting and measuring FHOC size, with specific metrics indicating performance.
• The study established a radiographic growth chart for FHOC in healthy Korean children.
• Automated measurements reduced inter- and intra-observer variability compared to manual methods.

Interpretation:

The DL-based approach provides a standardized, objective method for assessing FHOC size, enhancing clinical diagnosis and monitoring of pediatric hip conditions through improved reproducibility.

Limitations:

• The study was limited to healthy Korean children, which may affect generalizability and introduce population-specific biases.
• Only AP pelvic radiographs were used, excluding lateral views that may provide additional information.

Conclusion:

The developed DL algorithm offers a reliable tool for FHOC measurement, potentially improving pediatric hip joint assessments and providing essential reference values.