The diagnostic precision of computed tomography for traumatic cervical spine injury: An in vitro biomechanical investigation.

BACKGROUND: CT is considered the best method for vertebral fracture detection clinically, but its efficacy in laboratory studies is unknown. Therefore, our objective was to determine the sensitivity, precision, and specificity of high-resolution CT imaging compared to detailed anatomic dissection in an axial compression and lateral bending cervical spine biomechanical injury model. METHODS: 35 three-vertebra human cadaver cervical spine specimens were impacted in dynamic axial compression (0.5 m/s) at one of three lateral eccentricities (low 5% of the spine transverse diameter, middle 50%, high 150%) and two end conditions (19 constrained lateral translation and 16 unconstrained). All specimens were imaged using high resolution CT imaging (246 µm). Two clinicians (spine surgeon and neuroradiologist) diagnosed the vertebral fractures based on 34 discrete anatomical structures using both the CT images and anatomical dissection. FINDINGS: The sensitivity of CT was highest for fractures of the facet joint (59%) and vertebral endplate (57%), and was lowest for pedicle (13%) and lateral mass fractures (23%). The precision of CT was highest for spinous process fractures (83%) and lowest for pedicle (21%), uncinate process and lateral mass (both 23%) fractures. The specificity of CT exceeded 90% for all fractures. The Kappa value between the two reviewers was 0.52, indicating moderate agreement. INTERPRETATION: In this in vitro cervical spine injury model, high resolution CT scanning missed many fractures, notably those of the lateral mass and pedicle. This finding is potentially important clinically, as the integrity of these structures is important to clinical stability and surgical fixation planning.