查看文章

The geometric design of an implant significantly controls the success rates of implant surgery and influences osseointegration. The development of additive manufacturing (3D printing) also makes dental implants with complicated structures feasible to be manufactured. However, conventional computer-aided design (CAD) requires many engineering techniques, and it is time-consuming to generate a dental implant by its parametric design methodology. To more efficiently and conveniently generate the dental implant model, this study developed an Encoder–Decoder neural network with a multi-scale of images to be an alternative to a parametric implant generator in CAD. The network successfully generated the geometry of a dental implant by giving 15 geometrical parameters around 150 microseconds, which are the physical features to define a dental implant. In addition, users without any technical background …