Evaluation of manual and computerized methods for the determination of axial vertebral rotation

Abstract

Axial vertebral rotation is among the most important parameters for the evaluation of spinal deformities, and several manual and computerized methods have been proposed for its measurement. Routine manual measurement of axial vertebral rotation from three-dimensional (3D) images is error-prone due to the limitations of the observers, different properties of imaging techniques, variable characteristics of the observed anatomy, and difficulties in image navigation and representation. Computerized methods do not suffer from these limitations and may yield accurate results, however, they also require manual identification of multiple anatomical landmarks or neglect the sagittal and coronal inclinations of vertebrae. The variability of manual and computerized methods for measuring axial vertebral rotation in 3D images has not been thoroughly investigated yet. In this study we evaluated, compared and analyzed four different manual and a computerized method for measuring axial vertebral rotation. Using each method, three observers independently performed two series of manual measurements on 56 normal and scoliotic vertebrae in images, acquired by computed tomography (CT) and magnetic resonance (MR), which allowed the estimation of intra-observer, inter-observer and inter-method variability. The relatively low intra-observer standard deviation (0.8, 0.7 and 1.3 degrees for each observer), inter-observer standard deviation (1.3, 2.0 and 1.9 degrees for each observer pair), and inter-method standard deviation (best: 1.9 degrees) of the computerized method indicate that it is feasible for the determination of axial vertebral rotation and may represent an efficient alternative to manual methods in terms of repeatability, reliability and user effort.