Abstract

Treatment of a tooth that is seriously calcified, malposed, or fixed could make it challenging to decide the numberwhat’s more, position of openings on the floors of mash chambers. A novel method for locating root-canal orifices and pulp chambers is presented after analyzing pulp chambers from 3000 pulled teeth.

An essential but challenging step in dental surgical planning is the precise and automated segmentation of individual teeth and root canals from cone-beam computed tomography (CBCT) images. For efficient, precise, and fully automatic root canal segmentation from CBCT images, we propose a novel framework made up of two neural networks—DentalNet and PulpNet—in this paper. To begin, we use the proposed DentalNet to segment and identify tooth instances. After that, the affected tooth’s region of interest (ROI) is taken out and fed into the PulpNet for precise segmentation of the pulp chamber and root canal space. These two networks outperform a number of comparing  methods when tested on two clinical datasets and trained with multi-task feature learning. In addition, in order to enhance the surgical planning procedure, we incorporate our method into an effective clinical workflow. In two clinical case studies, our workflow effectively obtained the 3D model of the tooth and root canal for surgical planning in 2 minutes instead of 6 hours, resulting in satisfying outcomes in challenging root canal treatments.