A 2D-3D Registration Tool Capable of Accurately Quantifying Flexion- Extension Positioning in the Cervical Spine

Abstract

Flexion-extension x-rays are almost exclusively employed to analyze cervical spine kinematics. Yet, 2D radiographic measures of key vertebral metrics used to evaluate cervical stability are limited by x-ray source beam divergence, magnification errors and off-axis image acquisition. Three-dimensional CT images can be used to accurately measure these parameters, however flexion-extension CT scans are rarely acquired and do not provide information on the kinematics of the loaded cervical spine. This study evaluates the ability of an open source 2D-3D intensity-based image registration algorithm (xSePT) to create accurate flexion-extension 3D CT quality data from 2D flexion-extension radiographs and a single neutral 3D CT scan. Off axis 2D digitally rendered radiographs (DRRs) were generated from a set of 'gold standard' flexion-extension CT images of a single patient. The xSePT algorithm used the DRRs and a neutral CT to generate 3D flexion-extension CT images. Cervical vertebral metrics of subluxation, intervertebral disc height and interspinous process gap distance were compared between the 'gold standard' flexion-extension CT images, the 2D flexion-extension DRRs and the 3D output of the xSePT algorithm. The xSePT 2D-3D registration tool successfully aligned to the original flexion-extension CT data within 1° rotation and 0.5mm translation. The algorithm rapidly calculated values for the vertebral metrics equivalent to those based on the original flexion-extension CT data. Future evaluation of cervical pathology and kinematics under load may be possible through the application of this algorithm to generate 3D loaded flexion-extension data based on 2D standing flexion-extension x-rays and a single neutral unloaded CT scan.