|Ph.D Student||Yaron Holdstein|
|Subject||Geometric and Physical Modeling of Bone|
Micro-Structures as a Base for Scaffold Design
|Department||Department of Mechanical Engineering||Supervisor||Full Professor Fischer Anath|
|Full Thesis text|
This study focused on creating a structure upon which bone can grow in areas where the bone micro-structure is corrupted. This sort of structure is known as a scaffold, and it dictates how the bone grows. It is assumed that the resulting bone structure is influenced by the scaffold structure. Therefore the scaffold shape is critical from the point of view of functionality . The shape and topology of the scaffold structures that serve as the basis for bone implants have porous layouts. However, most standard structures are characterized by a simple and symmetric formation and thus are not customized to a specific patient, type of bone, or location inside the bone. Such standard structures may induce excessive stresses on the bone structure due to the sharp transition in the geometry between the real bone structure and the synthetic scaffold. These stresses may ultimately lead to scaffold failure. Hence, there is currently a gap between the currently available structure and the desired natural structure. We believe that our work will contribute considerably toward closing this gap .
Our aim in this research was to develop structural fillings, i.e. scaffolds, which vary in location, size and shape. Such fillings are irregular, but their textural behavior resembles that of their surroundings. Moreover, these fillings connect smoothly to the healthy neighborhood according to topological and geometrical attributes. Thus, their mechanical properties are topologically optimized with respect to that neighborhood .
In this study, two approaches were investigated: texture synthesis and semantic completion. Initially, the 2D texture synthesis method, which has shown promising results for 2D images of bone micro-structure, was extended to three dimensions. Two sub-approaches of texture synthesis were examined: pixel-by-pixel and block-by-block. The 3D extension of this approach showed good results for precise CAD models but was problematic for 3D irregular structural models. We then developed and tested the semantic completion approach of mapping healthy regions to the damaged regions. This approach includes smooth stitching between the regions. The advantage of this method is that the resulting scaffolds are irregular and have the same textural behavior as their surroundings. Thus, the mechanical behavior of the repaired bone is similar to that of healthy bone and exhibits no sharp changes in structure that might make the entire structure prone to fractures. To the best of our knowledge, such an algorithm has not yet been developed for 3D irregular micro-structures, and developing the method posed major challenges.