Reproducibility of measuring the shape and three-dimensional position of cervical vertebrae in upright position using the EOS stereoradiography system.

Abstract

An interobserver and intraobserver reproducibility study of the use of EOS stereoradiography system at the cervical spine. To investigate reproducibility of the determination of the vertebral shape, position, and orientation of C3-C7 vertebrae in vivo using the EOS stereoradiography system. Since CT and MRI 3-dimensional (3D) analysis of the spinal architecture are done in supine position, measurements of the relative position of the cervical vertebrae in vivo in standing position requires stereoradiography. The innovative EOS system is an accurate and promising tool for stereoradiography. Its reproducibility at the cervical spine had to be assessed. Twenty volunteers had biplanar radiographs of the cervical spine. C3-C7 were reconstructed by 2 observers, 2 times each. Each reconstruction was compared with the corresponding average object in term of shape (point-to-surface distance) and position (Tx, Ty, Tz, Rx, Ry, Rz in a local screw-frame). The 95% confidence interval of the error in shape was 1.83 mm. It was 0.84 mm, 1.42 mm, 0.58 mm, 2.53 degrees, 2.34 degrees, and 3.24 degrees for the position in Tx, Ty, Tz, Rx, Ry, and Rz. Intraobserver differences were not significant. Interobserver differences were significant for the shape and in Tx, Rx, and Ry (0.9 mm, 0.54 mm, 0.33 degrees, and 0.28 degrees). Overall reproducibility favorably compared with other imaging methods, whereas significant interobserver disagreements were narrow and partial.