Researchers from the University of Hong Kong used deep learning techniques to design dental crowns, which are caps placed on top of an existing tooth to restore and protect it.
Currently, most dental crowns are created using computer-aided design (CAD). The researchers say their methods generated aesthetically and functionally superior crowns compared to CAD methods.
Further, using AI to design crowns saves time and money and reduces the likelihood of human error.
How does it work?
The study, published in Dental Materials, used a dataset of 600 digital casts of second premolars and neighboring teeth. These were passed into a 3D neural network, a 3D-DCGAN, which learned various structural features from the healthy digital casts.
Once the model was trained, researchers used it to generate dental crowns and found it outperformed classically designed crowns across several variables.
Researchers say, “This study demonstrated that 3D-DCGAN could be utilised to design personalised dental crowns with high accuracy that can mimic both the morphology and biomechanics of natural teeth.”
Not only were the generated teeth predicted to be functionally superior to CAD-designed teeth, but they were also likely to outlast them.
The study has been well-received by other dentists. Tejas Patel, from Austin Cosmetic Dentistry in Texas, told Fox News that creating crowns with CAD “requires quite a bit of time for individual patients. This means a higher cost and more room for error.”
He also commented on the longevity of the AI-generated crowns, “With the use of generative AI, these processes can more accurately create personalized crowns using previous data and mold them with enough precision to last almost as long as real teeth.”
Patel says AI-generated crowns will be subject to clinical trials before rolling out to the public.
However, this is a quicker process in dentistry than in other medical fields, so patients could see the benefits within just a year or two.
3D neural networks in medicine
3D neural networking has been investigated for other medical applications, such as analyzing complex 3D diagnostic images produced by fMRI.
This technique has been used to classify brain hemorrhage, identify structural abnormalities indicative of Alzheimer’s disease, and classify gliomas, a type of brain tumor.
AI is supporting medical advances across virtually every field and discipline, and it won’t be long before AI-assisted techniques become the norm.
