水平集活动轮廓模型的3维牙齿重建

发布时间：2018-05-01 12:19

本文选题：牙齿分割 + CT图像　；参考：《中国图象图形学报》2016年08期

【摘要】：目的建立患者口腔每颗牙齿独立的3维模型对于计算机口腔修复的精确定位和量化评估具有十分重要的意义。针对口腔CT图像序列的牙齿形态变化和排列特点,提出一种新颖的基于水平集活动轮廓模型的3维牙齿重建方法。方法基于层间映射机制,对不同部位牙层切片采用不同的分割模型:利用先验形状约束能量、基于Flux模型的边缘梯度能量、基于先验灰度的局部区域能量相结合构造的单相混合水平集模型分割牙根层切片轮廓;利用结合区域竞争约束的双相混合水平集模型分割牙冠层切片轮廓;最后利用这些层间轮廓重建出牙齿3维模型。结果对不同位置的牙齿CT图像进行分割实验,结果表明,与现有的方法相比,本文方法具有较好的分割效果和较高的准确率,平均分割相似系数达到96%。结论本文的混合水平集模型能有效克服牙髓腔和牙槽骨的干扰以及图像灰度不均匀等问题,较准确地重建出每颗牙齿独立的3维模型,从而为制定口腔修复规划、生物力学分析等奠定坚实的基础。
[Abstract]:Objective to establish an independent 3D model of each tooth of patients with oral cavity is of great significance for accurate location and quantitative evaluation of computer dental repair. A novel 3D tooth reconstruction method based on the active contour model of horizontal set is proposed according to the characteristics of tooth morphology and arrangement in oral CT image sequence. Methods based on the mechanism of interlaminar mapping, different segmentation models were used for different sections of tooth layers: a priori shape was used to constrain energy, and the edge gradient energy based on Flux model was used. A single-phase mixed horizontal set model based on a priori gray scale combining local regional energy is used to segment the root slice contour, and a two-phase mixed horizontal set model combining regional competition constraints is used to segment the crown slice contour. Finally, the three-dimensional tooth model was reconstructed by using these interlaminar profiles. Results compared with the existing methods, this method has better segmentation effect and higher accuracy, and the average segmentation similarity coefficient is 96%. Conclusion the mixed level set model can effectively overcome the interference of pulp cavity and alveolar bone, and the image grayscale is uneven, so that the 3D model of each tooth can be reconstructed accurately, so as to make the dental restoration plan. Biomechanical analysis laid a solid foundation.
【作者单位】：嘉兴学院机电工程学院;
【基金】：浙江省自然科学基金项目(LQ14E050006) 国家自然科学基金项目(51405197)~~
【分类号】：R783;TP391.41

【相似文献】