Tooth shape reconstruction from dental CT images with the region-growing method.

Abstract

The three-dimensional shape information of teeth provides useful information. However, obtaining accurate three-dimensional shapes of teeth is difficult without extracting them physically. In this study, we aimed to develop a method for automatically extracting accurate three-dimensional shapes of teeth from dental CT images. The proposed method includes pre-processing and region extraction. Pre-processing is a combination of image-processing techniques that enhances tooth regions. In the region-extraction process, the region-growing method is introduced for extracting a region of each tooth. Constraint conditions determined by considering the characteristics of the structure of teeth are introduced for accurate extraction. Finally, morphological image processing is applied for eliminating discontinuous points. We carried out an experiment in which the three-dimensional shapes of teeth were reconstructed from dental CT images. Quantitative evaluation was performed by measuring the three-dimensional spatial accordance rates between the region obtained by the proposed method and the manually extracted region. The proposed method was significantly more accurate than an existing method at the 5% level. The experimental results showed that the proposed method reconstructs the shapes of teeth with high precision. However, an unextracted region remained at the surface of the enamel. Solving this problem and improving the extraction accuracy are important topics for future work.