Skip to Main Content
Exact information about the shape of a lumbar pedicle can increase operation accuracy and safety during computer-aided spinal fusion surgery, which requires extreme caution on the part of the surgeon, due to the complexity and delicacy of the procedure. In this paper, a robust framework for segmenting the lumbar pedicle in computed tomography (CT) images is presented. The framework that has been designed takes a CT image, which includes the lumbar pedicle as input, and provides the segmented lumbar pedicle in the form of 3-D voxel sets. This multistep approach begins with 2-D dynamic thresholding using local optimal thresholds, followed by procedures to recover the spine geometry in a high curvature environment. A subsequent canal reference determination using proposed thinning-based integrated cost is then performed. Based on the obtained segmented vertebra and canal reference, the edge of the spinal pedicle is segmented. This framework has been tested on 84 lumbar vertebrae of 19 patients requiring spinal fusion. It was successfully applied, resulting in an average success rate of 93.22% and a final mean error of 0.14 ± 0.05 mm. Precision errors were smaller than 1% for spine pedicle volumes. Intra- and interoperator precision errors were not significantly different.