High agreement of three-dimensional lower limb alignment analysis between a novel cone-beam CT and multislice CT

Lower limb alignment analysis based on three-dimensional (3D) models has become integral to preoperative planning. However, 3D models are typically reconstructed from data acquired in non-weight-bearing conditions. This study aims to validate a cone beam computed tomography (CBCT) for a three-dimensional assessment of the lower limb in upright position.

A fixed phantom limb, derived from supine CT data, was built and scanned five times by two different imaging modalities (CBCT and multislice computed tomography (MSCT)) in the upright and supine positions. A surface comparison between the 3D models generated from CBCT and MSCT was conducted. Furthermore, two observers performed alignment analysis by determining 31 landmarks on each 3D model. Intra- and interobserver variability of the landmarks and root mean square error (RMSE) of angles, axes, patellofemoral values, and leg lengths were evaluated.

The surface comparison revealed excellent agreement, with deviations primarily confined within the range of ±0.2 mm. Intra- and interobserver variability of all landmarks was less than 5 mm in both imaging modalities. The mean precision of all landmarks in CBCT and MSCT showed no statistically significant difference between both modalities (1.38 mm [SD 0.63] vs. 1.46 mm [SD 0.74], p = 0.444). For angles and axes, interobserver mean values in MSCT and CBCT differed by less than one degree in all measurements except femoral torsion (>2°). RMSE for angles and axes was less than one degree in most measurements except for tibial and femoral torsion.

CBCT and MSCT showed a high level of agreement in 3D imaging of the lower limb and assessing alignment and joint level metrics. By offering 3D imaging under physiological load, CBCT may have the potential to improve preoperative planning in clinical use.

N/A.