Abstract:The process of constructing precise geometry of human jaws from cone beam computed tomography (CBCT) scans is crucial for building finite element models and treatment planning. Despite the success of deep learning techniques, they struggle to accurately identify delicate features such as thin structures and gaps between the tooth-bone interfaces where periodontal ligament resides, especially when trained on limited data. Therefore, segmented geometries obtained through automated methods still require extensive manual adjustment to achieve a smooth and organic 3D geometry that is suitable for simulations. In this work, we require the model to provide anatomically correct segmentation of teeth and bones which preserves the space for the periodontal ligament layers. To accomplish the task with few accurate labels, we pre-train a modified MultiPlanar UNet as the backbone model using inferior segmentations, i.e., tooth-bone segmentation with no space in the tooth-bone interfaces, and fine-tune the model with a dedicated loss function over accurate delineations that considers the space. We demonstrate that our approach can produce proper tooth-bone segmentations with gap interfaces that are fit for simulations when applied to human jaw CBCT scans. Furthermore, we propose a marker-based watershed segmentation applied on the MultiPlanar UNet probability map to separate individual tooth. This has advantages when the segmentation task is challenged by common artifacts caused by restorative materials or similar intensities in the teeth-teeth interfaces in occurrence of crowded teeth phenomenon. Code and segmentation results are available at https://github.com/diku-dk/AutoJawSegment.

The Morphology-based Interactive Segmentation of Dental Models

[Oral Panorama Reconstruction Method Based on Pre-Segmentation and Bezier Function].

Tooth Segmentation on Dental Meshes Using Morphologic Skeleton.

Image Segmentation and Contour Correspondence and Branching Problem in Human Mandible Reconstruction

[Research on intelligent tooth segmentation method combining multiple seed region growth and boundary extension]

Single-Tooth Modeling for 3D Dental Model.

An automatic algorithm for repairing dental models based on contours

A Regional Energy Function Based Approach for Fast Tooth Segmentation from CBCT Images

A Hybrid Approach to Extracting Tooth Models from CT Volumes

Teeth segmentation using dental CT data

Model Adaptive Tooth Segmentation.

3D Tooth Mesh Segmentation with Simplified Mesh Cell Representation

Teeth-SEG: An Efficient Instance Segmentation Framework for Orthodontic Treatment based on Anthropic Prior Knowledge

3D human mandible reconstruction from CT data

3D Whole Tooth Model from CT Volume Using Thin-Plate Splines.

Clinically Applicable System For 3D Teeth Segmentation in Intraoral Scans using Deep Learning

Deformation Design Technology of Dental Restoration Model

A Semi-automatic Method for Detecting Dental Arch form in Triangular Mesh Model

Deep-Learning-Based Segmentation of Individual Tooth and Bone With Periodontal Ligament Interface Details for Simulation Purposes

Automatic Segmentation of Dental Cone Beam Computed Tomography Image Based on Level Set Method Using Morphology Operators and Polynomial Fitting

Research and Implementation for Tooth Surface Restoration Based on Membrane Deformation