Date of Graduation

12-2022

Document Type

Dissertation

Degree Name

Doctor of Philosophy in Materials Science & Engineering (PhD)

Degree Level

Graduate

Department

Materials Science & Engineering

Advisor/Mentor

Arun Nair

Committee Member

Matthew Leftwich

Second Committee Member

Uche Wejinya

Third Committee Member

Jingyi Chen

Fourth Committee Member

Panneer Selvam

Keywords

Bioinspired, Machine Learning, Tissue Scaffold, Ultrasound

Abstract

Porous scaffolds mechanically stimulated with ultrasound are used for the treatment of bone conditions such as osteoporosis or traumatic bone injury. However, one current issue in these treatment techniques is designing a scaffold structure with a high structural integrity that also does not impede tissue growth. A larger ultrasound wave transmission would also allow more of a scaffold to receive mechanical stimulation from ultrasound. Controlling mesenchymal stem cell seeding of tissue scaffolds also has potential for developing optimal seeding patterns to maximize or control tissue growth in the scaffold based on an individual’s physiology. There is also a need for the development of models to evaluate changes in bone pore structure, even when porosity is constant, to predict bone fracture risk more accurately to conditions such as osteoporosis, as well as developing models that can highlight the most important variables that affect the bone growth to tailor the scaffold design to optimize the required amount and rate of bone growth to meet patient’s physiological needs. This study develops experimental and computational models of the ultrasound wave transmission in a bioinspired scaffold structure with high structural integrity using 3D printed scaffolds, immersion ultrasound transducers, finite element simulations and the bioinspired structure is found to have equal amounts of bone tissue growth as structures with uniform pore shapes. A model of bone tissue growth in a bioinspired structure a bone tissue growth algorithm is then developed, and the bioinspired structure is found to have equal amounts of bone tissue growth as the structures with uniform pore shapes. The initial seeding of mesenchymal stem cells into the scaffold is also found to significantly affect the amount of bone tissue growth by as much as 120%. Lastly, unsupervised machine learning models evaluate the effects and presence of unknown patterns on bone tissue growth from different variables using machine learning. The unsupervised machine learning model of ultrasound wave transmission in the scaffolds developed in this study was found to be 81% accurate in classifying between an anisotropic and isotropic scaffold structure. The unsupervised machine learning model in this study classifying data clusters of bone and cartilage growth in the bioinspired and square pore scaffold structures compares the clusters to the effects of features such as pore geometry, scaffold material, ultrasound frequency and wave type, and mesenchymal stem cell distribution. This study thus demonstrates a bioinspired bone tissue scaffold structure with a high structural integrity to oppose structural failure while maintaining a high ultrasound wave transmission to allow more mechanical stimulation to grow bone tissue as well as develops models that can be used to design the optimal bone tissue scaffold design to meet a patient's specific physiological needs.

Available for download on Monday, February 17, 2025

Share

COinS