3D Shape Modeling Using Mesh Quality Preserved Deformable Models Based on Micro-CT Imaging

In this paper, a three-dimensional (3D) shape molding method was developed using deformable mesh models based on micro-CT imaging. The spatial constraint model took advantage of probabilistic segmentation to ensure that the deformable surface stayed in the proper configuration. Each seed point position obtained one binary segment of the micro-CT image, which was applied to construct the B-spline grid. A local neighborhood point was created around each initial position in the grid, which minimized the sum of the values of pixels behind the seed in the combined probability map. The benefit of our method is that the detail information was well-preserved, while the mesh quality was improved and execution time was reduced. Additionally, the accuracy was improved because of the shape refinement phase. A 3D human femur was successfully formed from micro-CT images, and the extensive experiments demonstrated the effectiveness and superiority of the proposed method.