3D Reconstruction of the Mandible from Panoramic Dental X-Ray Projections
- Panoramic radiography is a popular imaging technique in dental medicine. It omprehensively represents the entire jaw, including the mandible, maxilla, and teeth in a compact two-dimensional (2D) image. It offers several advantages over three-dimensional (3D) imaging, such as Cone Beam Computed Tomography (CBCT), including cost-effectiveness and reduced radiation exposure. However, its major drawback lies in its ambiguities in spatial distribution, leading to inaccuracies in linear measurements due to the dependency on the parameters used for generating the image. In this thesis, we propose a methodology for reconstructing a 3D anatomy of the mandible from a given 2D panoramic X-ray image. Our approach involves extracting information from the 2D scan, considering the projection geometry, and aiming to achieve a realistic spatial distribution. The methodology builds upon an existing intensity-based 2D-3D registration algorithm, which is modified to accommodate the unique geometric parameters of panoramic radiography. By employing a virtual X-ray setup, we generate a panoramic X-ray from a Statistical Shape and Intensity Model (SSIM) derived from 25 tetrahedral meshes. The intensity information for each mesh is extracted from CBCT scans representing jaw anatomies. To reconstruct the true mandible, the model is iteratively optimized by comparing the simulated X-ray with the original panoramic X-ray. We evaluate and establish the setup using synthetically generated panoramic X-rays from multiple mandibles obtained from CBCT scans. After analyzing the feasibility of this method, we conclude that 2D-3D registration for panoramic X-rays presents a challenging task, as the ambiguity of the 2D projection involved in panoramic X-ray imaging can result in varied 3D shape representations.
| Author: | Sandra Lukanek |
|---|---|
| Document Type: | Master's Thesis |
| Granting Institution: | Freie Universität Berlin |
| Advisor: | Stefan Zachow, Marko Leskovar |
| Date of final exam: | 2024/07/30 |
| Year of first publication: | 2024 |
| Page Number: | 49 |

