Description
The project "Realistic Neural Rendering of Dental Structures" aims to develop advanced techniques that leverage the capabilities of neural networks to achieve highly accurate and visually realistic digital representations of dental structures (unlike the rendering in the teaser image to the right :) ). The project will involve training neural networks (or using pre-trained networks) to render dental structures with high fidelity. The project will also explore the integration of styles, enabling dental professionals to interact with digital representations in a dynamic and intuitive manner.
By harnessing the power of deep learning algorithms, this project seeks to enhance the visualization of teeth. Rendering teeth realistically can be challenging due to their unique properties (such as translucency and subsurface scattering, complex surface details, etc.) Additionally, teeth exhibit a range of colors and textures, influenced by factors like age, hygiene, and other habits. Achieving accurate color reproduction and texture variations while considering the individual characteristics of each tooth is important.
The primary objective of this project is to improve patient communication. By generating lifelike renderings that closely resemble real-world dental structures, dental professionals will gain access to detailed and precise information, including enamel texture, occlusal surfaces, and anatomical variations. This enhanced visualization will enable them to, for example, effectively communicate treatment plans to patients.
Some (general) references:
1. https://onlinelibrary.wiley.com/doi/10.1111/cgf.13665
2. https://doi.org/10.1007/978-3-642-33191-6_33
3. https://pubmed.ncbi.nlm.nih.gov/31356110/
Requirements
- Interest and knowledge in rendering.
- Interest and/or knowledge in neural networks.
- Good programming skills.
- Creativity and enthusiasm.
Collaboration
This topic will be done in collaboration with ZAnGeSa GmbH, a company focusing on digital dentistry (https://www.zangesa.com/)