Session: 17-01-01: Research Posters

Paper Number: 139664

139664 - A Convolutional Neural Network Model for in Situ Characterization of Material Deposition Quality in Additive Fabrication of Bone Scaffolds With Complex Microstructure 

This research addresses the challenge of ensuring consistent extrusion quality in fabricated porous bone scaffolds, a critical factor influencing their structural integrity and biomedical functionality. The overarching problem involves the lack of automated and reliable methods for real-time assessment of extrusion quality during the manufacturing process. The long-term goal of this research is to establish a smart, reliable, and scalable quality control framework for additive fabrication of porous bone tissue scaffolds. The objective of the work is to develop a Convolutional Neural Network (CNN) model as an intelligent, automated prediction tool, adept at learning complex visual patterns of bone tissues indicative of different extrusion regimes. By integrating the developed CNN model into the manufacturing workflow, near real-time prediction and monitoring of extrusion quality can be achieved. The significance of this research extends to the broader field of tissue engineering, where the quality of fabricated scaffolds directly impacts their structural integrity and biological performance. The proposed model not only aids in identifying and rectifying manufacturing flaws promptly but also contributes to the overall reproducibility and reliability of scaffold fabrication for biomedical applications. In this work, the formation of a comprehensive dataset of fabricated scaffolds involves instances of (i) over extrusion, (ii) under extrusion, and (iii) normal extrusion regimes. The dataset consists of bone scaffolds fabricated by means of Pneumatic Micro-Extrusion (PME) additive manufacturing process (with a nozzle diameter, translation speed, layer height, and flow pressure of 860 µm, 10 mm/s, 680 µm, and 150 kPa, respectively) and composed of a synthesized, biocompatible composite material based on hydroxyapatite (HA), polysaccharide, an oxygen-generating material, and a ceramic component. The proposed CNN model is designed to effectively learn distinctive visual features from high-definition images captured by an integrated camera within the bioprinting system. The performance achieved through optimization of hyperparameters demonstrates the effectiveness of the CNN model in accurately predicting material extrusion morphology and quality. It was observed that the model was capable of predicting the extrusion regimes with an accuracy of ≥ 90%. Overall, the model proves to be robust across a range of extrusion scenarios, showcasing its potential as a practical and efficient solution for near real-time monitoring (and eventually control) of quality during the scaffold fabrication process. In addition, the proposed model offers a reliable, non-destructive means of identifying and categorizing extrusion issues. The results obtained in this study lay the foundation for the development of an automated quality control framework that can be seamlessly integrated into the scaffold fabrication workflow, providing near real-time feedback and aiding in achieving a robust and reproducible material extension regime for optimal fabrication of bone scaffolds. This contributes to the long-term goal of establishing standardized and efficient manufacturing processes for producing high-quality, patient-specific porous bone scaffolds in the realm of regenerative medicine.

Presenting Author: Roozbeh (Ross) Salary Marshall University (West Virginia State)

Presenting Author Biography: Dr. Roozbeh Ross Salary is an Assistant Professor of Mechanical and Biomedical Engineering in the College of Engineering and Computer Sciences at Marshall University. His current areas of research include Advanced Manufacturing, Biomanufacturing, Tissue Engineering, Machine Learning, and Artificial Intelligence (AI). Currently, Dr. Salary is serving as the director of the Lab for Advanced Manufacturing Engineering & Systems (LAMES) at Marshall University. Dr. Salary is a recipient of the College of Engineering Research Award as well as Pickens-Queen Teaching Award for excellence in both research and education at Marshall University.

Authors:

Ethan O'malley Marshall University
Roozbeh (Ross) Salary Marshall University (West Virginia State)