• Knackstedt, Mark Alexander
• Schmutz, Beat
• Herath, Buddhi
• Yarlagadda, Prasad Kdv

Abstract

INTRODUCTION <br/>The Voronoi structure is gaining popularity within the bone tissue engineering community to develop bone scaffolds, given its close resemblance to trabecular bone. Bone scaffold meshes designed for 3D printing should possess certain qualities such as being manifold and closed. This paper presents a semi-automated generative design workflow to design Voronoi bone scaffolds based on a CT imaged bone defect, including the fixation flanges if required. The output design is closed and manifold, which does not need any further mesh repairing operations, thereby making it readily 3D printable. <br/><br/>METHODS <br/>Aside from the image segmentation step, the core design workflow is set up in the software Rhinoceros 3D along with its visual programming module Grasshopper. Based on the OpenVDB [1, openvdb.org] library and the Grasshopper plugin Dendro (ecrlabs.com/dendro), a custom plugin was developed in-house for Grasshopper that uses signed distance fields to represent and manipulate a geometry implicitly. <br/>The DICOM data is segmented with a suitable medical image processing software, and the 3D reconstructed mesh of the bone defect is exported in STL format. This is imported into Rhinoceros 3D and by using Boolean operations, SUB-D functions and the in-house developed plugin, a solid geometry of the scaffold region including the fixation flanges is constructed manually (Figure 1.1). Within Rhinoceros 3D, “Surface” geometries are created to demarcate the fixation flanges that are to remain solid (Figure 1.2) and are assigned to the respective node in the Grasshopper workflow, which will be used to split the mesh. Once the number of Voronoi seed points is selected using a number slider, they will be randomly distributed within the region (Figure 1.3) and the geometry will be partitioned into Voronoi cells (Figure 1.4). The edges of these cells and the curve intersections of the split mesh are isolated (Figure 1.5), and a level set is wrapped around them to form a solid lattice. Using implicit Boolean operations, the pores of the scaffold (Figure 1.6) and the porous lattice (Figure 1.7) and are created, converted back to a closed manifold mesh and 3D printed (Figure 1.8). Finally, the porosity, surface area of the scaffold and the effective pore diameters are computed. <br/><br/>RESULTS AND DISCUSSION <br/>The resultant mesh is a closed manifold mesh that does not need any mesh repairing operations prior to 3D printing. The implicit Boolean operations are computationally fast and robust. Once setup, the workflow functions as a semi-automated process where any variable such as the strut diameter, number of seed points, demarcations of the fixation flanges, scaffold geometry, etc. can be changed by the user and the generative workflow will regenerate the design accordingly.

Topics

• porous
• impedance spectroscopy
• pore
• surface
• porosity
• level set