CBCT image segmentation of tooth-root canal based on improved level set algorithm
| Autor: | Zhao Qun-fei, Tang Zi-sheng, Xia Wenjun, Yu Zichun |
|---|---|
| Rok vydání: | 2020 |
| Předmět: |
business.industry
Computer science Root canal 3D reconstruction Pattern recognition 02 engineering and technology Function (mathematics) Image segmentation Image (mathematics) Set (abstract data type) 03 medical and health sciences 0302 clinical medicine medicine.anatomical_structure Level set stomatognathic system 0202 electrical engineering electronic engineering information engineering medicine 020201 artificial intelligence & image processing Segmentation Artificial intelligence business 030217 neurology & neurosurgery |
| Zdroj: | CIPAE |
| DOI: | 10.1145/3419635.3419654 |
| Popis: | Image segmentation of the root canal is important basis for the establishment of a three-dimensional model for application value clinical diagnosis and education. A total of 1554 CBCT images of polymorphic roots of 10 in vivo and 8 in vitro teeth were preprocessed by the adaptive enhancement algorithm of CLAHS and Laplace-g, respectively. An improved level set algorithm was then used to segment the tooth-root canal image. In the process of curve evolution and convergence of the improved level set algorithm, the evolution of tooth-root canal contour was constrained by adding a new regularization function. Based on the similarity of root canal contours between adjacent sections of a tooth, a self-qualifying method was established to determine the initial contour of the root canal. In addition, a process was set up to analyze the data of a neighboring tooth to improve the segmentation process. Using the stated improvements, a set of root canal image segmentation methods was established for a single tooth based on the improved level set algorithm. Experimental results show that the average accuracy of the improved level set algorithm for root canal image segmentation is 84.7%, while in clinical trials, the average qualified rate is 90.4%. |
| Databáze: | OpenAIRE |
| Externí odkaz: |
|