An effective 3-D visualization system of spine MR images

The three dimensional (3-D) visualization of spine MR images can provide abundant and accurate information in spine diseases diagnosing. In this paper we propose a 3-D visualization system for vertebra and spine. We firstly segment the individual vertebra from spine MR images, and then the ray casting algorithm is employed for vertebra and spine 3-D visualization. The intensity inhomogeneity in MR images makes the segmentation of vertebra a big challenge. We combine the level set method with a knowledge-based disks detection method to resolve the problem. Experiments based on spine MR images provided by College of Medicine, University of Cincinnati, prove the feasibility and efficiency, which show that the proposed system can reconstruct 3-D vertebra with clear 3-D profile, and the reconstructed 3-D spine could reflect curvature of the spine.