Vertebral cortical thickness and cortical bone density: an automated CT assessment - towards enhanced spine segmentation

Abstract

In this paper a non-invasive method using routine CT scan to assess the vertebral geometry through normalised Cortical Thickness (CTh) and Cortical Bone Density (CBD) is proposed. This paper aims to propose a new automated method to segment cortical bone and measure its thickness and local density. This method were then used as a tool to compare these parameters between different vertebra models (in-vivo, cadaver and swine) and vertebra levels. An automated technique to segment cortical bone was proposed, assuming a two Gaussian bone density distribution. 42 vertebrae (3 high-thoracic, 3 low-thoracic and 1 lumbar vertebra for each subject) from three sub-groups (human in-vivo, cadaver and swine) were investigated. In the human in-vivo sub-group, the vertebral level was shown to influence normalised CTh and CBD. The CBD was found uniform within all the functional areas of the vertebral body (p > 0.05), while the normalised CTh showed significant differences (p < 0.001). Both CBD and normalised CTh showed significantly different between the inferior articular processes area and the posterior arch area d (p < 0.02 *). In each of the three sub-groups (human in-vivo Vs cadaveric Vs swine), normalised CTh and CBD were found significantly different across most functional areas (p < 0.001 ***). The proposed method offers an automated and accurate way of measuring cortical thickness and cortical bone density. Vertebra level and vertebra function areas were found to have influences on both proposed cortical bone characteristics. The influence of the vertebral level and of the vertebral functional areas on normalised CTh and CBD were reported within in-vivo and two vertebral models. Such a methodology could be used as a tool for image-guided surgery.

Keywords

• Segmentation
• cortical
• bone
• thickness
• vertebra