What is the association between MRI and conventional radiography in measuring femoral head migration?

Background and purpose — Pelvic radiographs are traditionally used for assessing femoral head migration in residual acetabular dysplasia (RAD). Knowledge of the heightened importance of cartilaginous structures in this condition has led to increased use of MRI in assessing both osseous and cartilaginous structures of the pediatric hip. Therefore, we assessed the relationship between migration percentages (MP) found on MRI and conventional radiographs. Second, we analyzed the reliability of MP in MRI and radiographs. Patients and methods — We retrospectively identified 16 patients (mean age 5 years [2–8], 14 girls), examined for RAD during a period of 2½ years. 4 raters performed blinded repeated measurements of osseous migration percentage (MP) and cartilaginous migration percentage (CMP) in MRI and radiographs. Pelvic rotation and tilt indices were measured in radiographs. Bland–Altman (B–A) plots and intraclass correlation coefficients (ICC) were calculated for agreement and reliability. Results — B–A plots for MPR and MPMRI produced a mean difference of 6.4 with limits of agreement –11 to 24, with higher disagreements at low average MP values. Mean MPR differed from mean MPMRI (17% versus 23%, p < 0.001). MPR had the best interrater reliability with an ICC of 0.92 (0.86–0.96), compared with MPMRI and CMP with ICC values of 0.61 (0.45–0.70) and 0.52 (0.26–0.69), respectively. Intrarater reliability for MPR, MPMRI and CMP all had ICC values above 0.75 and did not differ statistically significantly. Differences inMPMRI and MPR showed no correlation to pelvic rotation index, pelvic tilt index, or interval between radiograph and MRI exams. Interpretation — Pelvic radiographs underestimated MP when compared with pelvic MRI. We propose CMP as a new imaging measurement, and conclude that it has good intrarater reliability but moderate interrater reliability. Measurement of MP in radiographs and MRI had mediocre to excellent reliability