Comparative Study of Hip MRI and Hip Arthrography in Localizing the Femoral Head Center in ChildrenUnder 5 Years Old

BACKGROUND: Locating the center of the nonossified femoral head on anteroposterior (AP) pelvic radiographs in children under 5 remains challenging. Hip arthrography serves as an intraoperative clinical reference in our setting, but is invasive. The agreement of magnetic resonance imaging (MRI) using bony landmarks with arthrography has not been fully established. METHODS: Children under 5 years old with contemporaneous hip arthrograms, hip MRI, and AP pelvic radiographs were included in this retrospective study. The femoral head center was determined by the Mose concentric circle method on arthrography and MRI and proportionally projected onto a predefined coordinate system on AP pelvic radiographs. Agreement between methods for X and Y coordinates and the center-edge angle (CEA) was assessed using intraclass correlation coefficients (ICC). Intraobserver and interobserver reliability for localization and CEA measurement was assessed in a subset. In normal hips, the age-related distribution of the femoral head center was evaluated with Spearman correlation and linear regression. RESULTS: Forty-two cases (84 hips) were analyzed; 45 normal hips (54%), median age 18 months (range: 1 to 60). In normal hips, MRI showed poor agreement with arthrography for the X coordinate (ICC=0.13), good agreement for the Y coordinate (ICC=0.89), and good agreement for CEA (ICC=0.89). Reliability was good to excellent (intraobserver ICC: 0.82 to 0.91; interobserver ICC: 0.78 to 0.89). Using arthrography-defined the femoral head centers in normal hips, the mean X-coordinate was 0%±2% of femoral neck width and showed no correlation with age (r=0.17, P=0.27). The Y coordinate showed a strong negative correlation with age (r=-0.90, P