Design of a spinal cord phantom

Spinal cord plays an important role in human daily life and its lesion has usually dramatic consequences. A key point in spinal cord injury either for research or treatment is the localization of the injury. None of the existing technique allows a precise identification of the functional segment of the spinal cord which is injured. Many previous works designed spinal cord phantoms which however do not directly partition the spinal cord into its functional segments. This master thesis addresses this challenge by providing clinicians, neurosurgeons and researchers with a model which will directly partition the spinal cord and thus which will help to estimate the position of its functional segments. We designed a spinal cord phantom using standard clinical practice MR images of spinal cord and the proportions of spinal cord segments based on post-mortem anatomical studies. Semi automatic image segmentation based on region growing algorithm allows to process sagittal and axial MR images in order to construct a 3D volume of the spinal cord. Middle line of the volume has been inferred and partitioned thanks to 3D curve length computation and spline interpolation. This finally resulted after projection on 2D plan to the partitioning of the spinal cord in all sagittal plans crossing it.