Measurement of vertebral rotation in adolescent idiopathic scoliosis using three-dimensional magnetic resonance imaging.

Abstract

This report examines a technique for measurement of axial vertebral rotation using magnetic resonance imaging. To assess the reproducibility of three-dimensional magnetic resonance imaging in the measurement of vertebral rotation at individual endplates in patients with adolescent idiopathic scoliosis. Deformity in the sagittal and coronal planes in patients with adolescent idiopathic scoliosis can be readily assessed from plain radiographs, but the degree of deformity in the axial plane is more difficult to determine. Plain film techniques have inherent inaccuracies because of loss of definition of anatomic landmarks, and the use of computed tomography is limited by the high radiation dose associated. Magnetic resonance imaging provides a means of imaging scoliotic deformity that allows multiplanar reconstruction and that involves no use of ionizing radiation. Ten patients with adolescent idiopathic scoliosis were imaged in a Siemens 1-Tesla impact scanner. Three-dimensional volume images of the apical five vertebrae were obtained in the axial plane and were postprocessed through multiplanar reconstruction. Sections through the superior and inferior endplates of each vertebra were selected in the sagittal and coronal planes, allowing axial reconstructions to be obtained in the plane of each endplate. Vertebral rotation was measured by identifying datum points on the inner surfaces and at the junction of the laminas and comparing the angle subtended by these points with a vertical drawn by the computer. Measurements were obtained from the single scanning sequence on two occasions by one observer and on one occasion by a second observer. Interobserver and intraobserver error was evaluated and correlation with readings obtained from plain films using Perdriolle's torsiometer method assessed. The interobserver variation had a mean of 3.02 degrees (range, 0-10 degrees) and a 95% confidence interval of [2.51 degrees, 3.53 degrees]. The intraobserver variation had a mean of 2.56 degrees (range, 0-7 degrees) and a 95% confidence interval of [1.83 degrees, 3.29 degrees]. The mean difference between measurements obtained from magnetic resonance imaging and plain film was 3.29 degrees (range, 0-12 degrees) with a 95% confidence interval of [1.43 degrees, 5.15 degrees]. The degree of vertebral rotation can be accurately and reproducibly assessed by three-dimensional magnetic resonance imaging. Measurements can be made through individual endplates that allow assessment of the relative amount of intervertebral and intravertebral deformity.