Reliability of Low-dose Biplanar Radiography in Assessing Pediatric Torsional Pathology

Background: Low-dose biplanar radiographs (LDBRs) significantly reduce ionizing radiation exposure and may be of use in evaluating lower extremity torsion in children. In this study, we evaluated how well femoral and tibial torsional profiles obtained by LDBR correspond with 3-dimensional (3D) computed tomography (CT) and magnetic resonance axial imaging (MRI) in pediatric patients with suspected rotational abnormalities. Methods: Patients who had both LDBR and CT/MRI studies performed for suspected lower extremity rotational deformities were included. Unlike previous publications, this study focused on patients with lower extremity torsional pathology, and bilateral lower extremities of 17 patients were included. CT/MRI torsion was measured using the Reikerås method, after conversion to 3D reconstructions. The LDBRs were deidentified and sent to the software division of EOS imaging, who created 3D reconstructions and evaluated each reconstruction for the torsional quantification of the femurs and tibiae. These imaging modalities were compared using correlation statistics and Bland-Altman analyses. Results: The mean age of the cohort was 12.1±1.7 years old. Torsional values of the femur were significantly lower in LDBRs versus 3D CT/MRIs at 17.7±15.1 and 23.3±17.3, respectively (P=0.001). Torsional values of the tibia were similar in LDBRs versus 3D CT/MRIs at 23.6±10.6 and 25.3±11.2, respectively (P=0.503). There was a good intermodality agreement between LDBR and 3D CT/MRI torsional values in the femur (intraclass correlation coefficient=0.807) and tibia (intraclass correlation coefficient=0.768). Bland-Altman analyses showed a fixed bias with a mean difference of −5.6±8.8 degrees between femoral torsion measurements in LDBRs versus 3D CT/MRIs (P=0.001); 15% (5/34) of femurs had a clinically significant measurement discrepancy. Fixed bias for LDBR measurements compared with 3D CT/MRIs for the tibia was not observed (P=0.193), however, 12% (4/34) of tibias had a clinically significant measurement discrepancy. Conclusion: Although we found strong correlations between torsional values of the femur and tibia measured from LDBRs and 3D CT/MRIs, torsional values of the femur produced from LDBRs were significantly lower than values obtained from 3D CT/MRIs with some notable outliers. Level of Evidence: Level III.