The development of automatic cranial implant design software using AI algorithms

Abstract: 

In the medical history, reconstruction of a cranial implant design is a challenge. Open-source CAD design software was used to design Cranial implants manually. This process was very costly and time consumable for patients and doctors.

In this study, automatic software is developed using deep learning algorithms to design and reconstruct a cranial implants automatically in the short notice of time. Deep Neural Network (DNN) was trained on various datasets including 200+ defected and completed models to predict the implants directly from the defected skull. The output observed was quite good to accept the reconstructed implant using this software. This algorithm generates high quality implant in two steps. In step one; an encoder-decoder detects the boundary lines of a defected skull to learn the actual size, shape and other parameters of the skull. Secondly, after the training and learning, implant is generated according to the required shape and size of the defect.