Three-dimensional assessment of lower limb alignment: Accuracy and reliability

Introduction: Three-dimensional (3D) surgical planning and patient-specific implants are becoming increasingly popular in orthopedics and trauma surgery. In contrast to the established and standardized alignment assessment on two-dimensional (2D) long standing radiographs (LSRs) there is neither a standardized nor a validated protocol for the analysis of 3D bone models of the lower limb. This study aimed to create a prerequisite for pre-operative planning. Methods: According to 2D analysis and after meticulous research, 24 landmarks were defined on 3D bone models obtained from computed axial tomography (CT) scans for a 3D alignment assessment. Three observers with different experience levels performed the test three different times on three specimens. Intraobserver and interobserver variability of the landmarks and the intraclass correlation coefficient (ICC) of the resulting axes and joint angles were evaluated. Results: Overall, the intraobserver and interobserver variability was low, with a mean deviation <5 mm for all landmarks. The ICC of all joint angles and axis deviations was >0.8, except for tibial torsion (ICC = 0.69). All knee joint angles showed excellent ICC (>0.95). Conclusions: Using the defined landmarks, a standardized 3D alignment assessment with low intraobserver and interobserver variability and high ICC values for the knee joint angles can be performed regardless of examiner's experience. The described method serves as a reliable standardized protocol for a 3D malalignment test of the lower limb. Three-dimensional pre-operative analysis might enhance understanding of deformities and lead to a new focus in surgical planning. (C) 2018 Elsevier B.V. All rights reserved.