Paper #59 - ISHA Annual Scientific Meeting 2016

Accuracy and Reliability of Computed Tomography and Magnetic Resonance Imaging Compared to True Anatomic Femoral Torsion

James D. Wylie, MD, MHS, Salt Lake City, UT UNITED STATES
Michael J. Beebe, MD, Memphis, TN UNITED STATES
Garrett Bodine, BS, Salt Lake City, UT UNITED STATES
Ashley Lynn Kapron, PhD, Salt Lake City, UT UNITED STATES
Travis Maak, MD, Salt Lake City, UT UNITED STATES
Omer Mei-Dan, MD, Aurora, CO UNITED STATES
Stephen Kenji Aoki, MD, Salt Lake City, UT UNITED STATES

University of Utah, Salt Lake City, UT, UNITED STATES

FDA Status Not Applicable

Summary: Axial computed tomography and magnetic resonance imaging are the most accurate way to measure femoral version and should be used in clinical practice.

Abnormal torsion of the femur has been previously correlated to lower extremity pathologies. While computed tomography(CT) evaluation of femoral torsion is the gold standard, magnetic resonance imaging (MRI) has been proposed as a viable alternative, avoiding the need for radiation exposure. The aim of this project was to determine the accuracy and consistency of four previously described MRI and CT femur rotational series in comparison to the true anatomic torsion of cadaveric femurs.

Twelve cadaveric femora were stripped of soft tissue. Four advanced imaging series were obtained for each specimen: CT with axial cuts of the femoral neck (CT-axial); CT with oblique cuts in line with the femoral neck (CT-oblique); MRI with axial cuts of the femoral neck (MR-axial); MRI with oblique cuts in line with the femoral neck (MR-oblique). To assess true torsion, anatomic specimens were placed with the posterior femoral condyles flat on a dissection table and a single lens reflex digital camera captured a direct axial image (0° of abduction) of each femur after which lens aberration was corrected. Three independent reviewers performed all measurements, including true torsion, using imaging software.
Bland Altman analysis was repeated with the data from each reviewer

Inter-observer repeatability for all groups was high at 0.95, 0.87, 0.90, 0.97, and 0.92 for CT-axial, CT-oblique, MR-axial, MR-oblique, and true torsion, respectively. When compared to true anatomic torsion of the specimen, CT-axial had the lowest mean difference for all three observers (all <1°) and held the tightest 95% limits of agreement for 2/3 observers. CT-oblique and MR-axial showed mean differences less than 2 degrees for all three observers. MR-oblique showed the greatest mean difference compared to the true anatomic torsion for all three observers (3.4-4.7°) and the largest 95% limits of agreement. As torsion increased from neutral into ante torsion, MR-oblique linearly overestimated the rotation compared to true anatomic torsion.

While both MR-axial and CT-oblique were accurate to a level that is likely less than clinical significance, CT-axial was both most accurate and consistent when compared to the true torsion of the femur and should thus be considered the gold standard imaging modality. MR-oblique imaging protocols can overestimate true version and showed the greatest variation and therefore should not be used.