| About this Abstract |
| Meeting |
2021 TMS Annual Meeting & Exhibition
|
| Symposium
|
Advances in Biomaterials for 3D Printing of Scaffolds and Tissues
|
| Presentation Title |
Design and Evaluations System for 3D-printed Dental Implants Based on Deep Neural Networks |
| Author(s) |
Pei-Ching Kung, Chai-Wei Hsu, An-Cheng Yang, Nan-Yow Chen, Nien-Ti Tsou |
| On-Site Speaker (Planned) |
Pei-Ching Kung |
| Abstract Scope |
In this study, a framework of deep neural networks for the custom design and evaluation of 3D-printed dental implants was proposed. The network consisted of three parts: Firstly, the evaluator successfully substitutes the mechano-regulatory method, which is one of the promising algorithms to numerically predict the corresponding implant performance indexes. High correlation coefficients were achieved for the evaluator part (greater than 0.94). Secondly, the designer well predicts the possible geometric features by giving the expected performance indexes, where high correlation coefficients can also be achieved. Finally, the generator builds a corresponding implant based on the geometric features predicted by the designer. The IoU (Intersection over Union) score of 0.997 for the generator part was also obtained. The current work provides a proof of concept that deep learning approaches can substitute the time-dependent, highly-complex, and multi-physical models/theories. Moreover, it provides an in-depth understanding of implants design with no prior professional knowledge. |
| Proceedings Inclusion? |
Planned: |