The development of a novel AI Hip Growth Chart represents a significant step forward in pediatric radiology. A recent study introduced a deep learning (DL)-based algorithm for automatically measuring the size of the femoral head ossification center (FHOC) in children. Historically, manual measurement of the FHOC on radiographs is often subjective and prone to inter-observer variability, which limits the standardization of growth assessment. This new DL-based approach aims to provide an objective and reproducible reference, which is crucial for the early detection of pediatric hip abnormalities.
Deep Learning for Automated FHOC Measurement
Researchers developed a three-stage deep learning algorithm to ensure accuracy and consistency. The process begins with detecting the region-of-interest, followed by precise FHOC segmentation, and finally, a landmark-based size computation. Agreement between the AI-derived measurements and those taken manually by experienced radiologists was exceptionally close. Consequently, the mean differences were kept within ±0.5 mm, which is a strong indicator of the algorithm’s precision. Furthermore, the overall agreement was supported by high concordance correlation coefficients and consistently low error metrics, demonstrating clinical viability. Because of this high performance, the AI system can effectively standardize a measurement previously known for its inherent subjectivity.
Clinical Relevance of the AI Hip Growth Chart
Establishing standardized reference ranges for pediatric skeletal development is paramount in orthopedics. The AI-derived measurements were utilized to construct FHOC growth curves using quantile polynomial regression. The resulting growth curves demonstrated strong predictive accuracy, with adjusted R² values exceeding 0.92 for both female and male cohorts. Moreover, these growth charts established reliable reference percentiles, ranging from the 5th to the 95th, providing clinicians with objective, age-specific FHOC size data. This is particularly relevant for conditions like developmental dysplasia of the hip (DDH) and other growth-related pathologies. For instance, the use of deep learning models for measuring various hip parameters has previously been shown to yield accurate and objective results, helping to identify hip dysplasia and femoroacetabular impingement. Therefore, the standardized AI-derived growth chart offers a robust tool for assessing normal versus abnormal development, potentially improving diagnostic consistency in clinical practice.
Frequently Asked Questions
Q1: What is the femoral head ossification center (FHOC)?
The FHOC is a critical anatomical structure in pediatric hip development. It is the center where bone begins to form in the cartilaginous head of the femur, and its size and timing of appearance are important radiographic indicators of a child’s skeletal maturity and hip health.
Q2: How does a deep learning algorithm improve FHOC measurement?
The deep learning algorithm automates the measurement process on pelvic radiographs, eliminating the human subjectivity and variability associated with manual measurements. The system is designed to provide highly reproducible, precise, and objective size measurements, which ensures a standardized assessment across all patients. Since high agreement with expert radiologist measurements was shown, the AI is considered a reliable tool.
References
- Lee BD et al. Deep learning-based automatic measurement of the femoral head ossification center in healthy Korean children: development of a novel radiographic growth chart. Eur Radiol. 2026 Jan 13. doi: 10.1007/s00330-025-12263-z. PMID: 41528475.
- Deep Learning for Fully Automated Radiographic Measurements of the Pelvis and Hip.
- A Cam Morphology Develops in the Early Phase of the Final Growth Spurt in Adolescent Ice Hockey Players: Results of a Prospective MRI-based Study.
- Revised IAP growth charts for height, weight and body mass index for 5- to 18-year-old Indian children.