Femoral and tibial torsion measurements in children and adolescents: comparison of MRI and 3D models based on low-dose biplanar radiographs.

Abstract

The aim of our study was to evaluate the reliability and interchangeability of femoral (FT) and tibial torsion (TT) measurements in children using magnetic resonance (MR) imaging compared to measurements on 3D models based on biplanar radiographs (BPR). FT and TT were measured in 60 children (mean age 10.1years; range 6.2-16.2years; 28 female) using axial MR images by two readers. MR measurements were compared to measurements based on BPR-3D models by two separate independent readers. Interreader and intermethod agreements were calculated using descriptive statistics, the intraclass correlation coefficient (ICC), and Bland-Altman analysis. FT/TT was -8.4°-54.1°/0°-45.9° on MR images and -13°-63°/4°-52° for measurements on BPR-3D models. The median of difference between the two methods was -0.18° (range -13.6°-19.1°) for FT and -0.20° (range -18.4°-9.5°) for TT, respectively. Interreader agreement (ICC) of FT/TT measurements was 0.98/0.96 on MR images and 0.98/0.94 on BPR 3D models. Intermethod agreement (ICC) for MR measurements was 0.95 [95% confidence interval (CI), 0.93-0.96] for FT and of 0.86 (CI, 0.24-0.95) for TT. Mean interreader differences at MR were 3.1° (0.0°-8.0°) for FT and 3.2° (0.1°-9.5°) for TT. On Bland-Altman plots all measurements were within the 95% limit of agreement (-10.8°; 11.5° for FT; -14.6°; 4.2°) for TT-except for five measurements of FT and six measurements of TT. FT measurements on MR images are comparable to measurements using BPR-3D models. TT measurements differ between the two modalities, but the discrepancy is comparable to measurement variations between CT and BPR.