Teeth and Jaw Segmentation from CBCT images Using 3D Deep Learning Models

Abstract

Abstract—the outcome of oral surgery is dependent on precise preoperative planning. In order to properly plan for dental surgery or suitable implant placement, it is necessary for accurate segmentation of the jaw tissues: the teeth, the cortical bone, the trabecular core, and overall, the inferior alveolar nerve. Achieving 3D virtual surgical planning is very hard and is a major drawback because most of the existing jaw segmentation methods require a large amount of expert knowledge for manual or partially automatic segmentation. What is more, due to the lack of experienced doctors and experts, high-quality expert knowledge is hard to achieve in practice. Furthermore, metal artifacts and large variations in their shape and size among individuals influence the segmentation of the jaw in CBCT scans seriously. Therefore, this paper presents a new automatic jaw segmentation in CBCT scans dataset. This approach is divided into two main phases; the first phase adopts the use of image processing techniques to extract the region of interest (ROI) which is the area containing the jaws. It is an important phase as it prepares the 3D image data before using it in the second phase as it makes the training more efficient rather than training the model overall 3D head of CBCT scans. In the second phase, we use several models to segment each jaw separately and then gets a 3D segmentation view of both jaws. We implement such a segmentation approach on head CBCT scans given as Dicom Numpy files datasets and then evaluate the performance. Our proposed approach has been tested in a validation set that has been evaluated using Dice’s coefficient and the Attention Unet model achieved promising results with a 91