Correction for patient sway in radiographic biplanar imaging for three-dimensional reconstruction of the spine: in vitro study of a new method.

Background: Three-dimensional (3D) reconstructions of the spine in the upright position are classically obtained using two-dimensional, non-simultaneous radiographic imaging. However, a subject's sway between exposures induces inaccuracy in the 3D reconstructions.

Purpose: To evaluate the impact of patient sway between successive radiographic exposures, and to test if 3D reconstruction accuracy can be improved by a corrective method with simultaneous Moire-X-ray imaging.

Material And Methods: Using a calibrated deformable phantom perceptible by both techniques (Moire and X-ray), the 3D positional and rotational vertebral data from 3D reconstructions with and without the corrective procedure were compared to the corresponding data of computed tomography (CT) scans, considered as a reference. All were expressed in the global axis system, as defined by the Scoliosis Research Society.

Results: When a sagittal sway of 10 degrees occurred between successive biplanar X-rays, the accuracy of the 3D reconstruction without correction was 8.8 mm for the anteroposterior vertebral locations and 6.4 degrees for the sagittal orientations. When the corrective method was applied, the accuracy was improved to 1.3 mm and 1.5 degrees , respectively.

Conclusion: 3D accuracy improved significantly by using the corrective method, whatever the subject's sway. This technique is reliable for clinical appraisal of the spine, if the subject's sway does not exceed 10 degrees . For greater sway, improvement persists, but a risk of lack of accuracy exists.