A new computer-aided technique for analysis of lateral cervical radiographs in postoperative patients with degenerative disease.

Abstract

Radiographs were measured by four independent observers and remeasured by three of the observers. To assess the reliability of a new computer-aided measurement technique. Many studies have reported sagittal-plane distance and angle measurements in the cervical spine. Common measurement methods involve manual line drawing on lateral radiographs with manual or computer-aided distance and angle computation. In patients with anterior cervical fusion, changes in bony geometry could present difficulties for many existing methods. Digitized lateral cervical radiographs are imported into a graphics software package. Outlines of the vertebral bodies and spinous processes are traced on the best-quality film and transformed to match the bony geometry on each remaining radiograph from the same patient. Intervertebral distance and angulation are calculated from centers of mass of the outlined elements. Segmental measurements were collected for 27 lateral radiographs from nine patients with anterior cervical discectomy and fusion. Intraclass and interclass correlation coefficients were calculated and used to compute standard errors of measurement. High intraclass and interclass correlations (ICCs) and low measurement errors were calculated for both distance and angle measurements. Intraexaminer mean ICCs were 0.92 for interbody distance and 0.93 for segmental angle, with standard errors of measurement (SEMs) of 3.26% interbody distance (approximately 0.65 mm) and 1.20 degrees sagittal-plane rotation. Mean interexaminer ICCs were 0.91 for interbody distance and 0.86 for segmental angle, with SEMs of 3.58% interbody distance (approximately 0.72 mm) and 1.77 degrees sagittal-plane rotation. The measurement method is reliable for both interbody distance and segmental angles within and among examiners. Whereas many existing measurement methods require normal radiographs to locate specific anatomic points, given intact spinous processes, the present method functions even with various radiographic abnormalities and in the presence of surgical decompression, degenerative disease, and cervical hardware. Because it does not rely on specific anatomic points, the present method is robust with respect to changes in the bony anatomy over time.