基于颌面部CT图像的下颌骨自动分割与三维重建方法

Automatic segmentation and three-dimensional reconstruction of mandible from maxillofacial CT image

提出一种下颌骨部分的自动提取与重建的方法,对颌面部CT图像使用U-Net神经网络自动分割出下颌骨部分,在移动立方体算法基础之上提出融合MCRC算法对分割结果进行三维模型重建,最后使用平滑处理优化模型表面。结果显示,该方法提高了实时交互绘制的能力,减少医生手动标注的工作量,可以有效减少三角面片的产生,改善模型显示效果,提高模型的结构清晰度与真实感,具有一定的实用价值。

An automatic extraction and reconstruction method of the mandible is proposed.The mandible is automatically segmented from the maxillofacial CT image using U-Net.Based on the marching cubes algorithm,a fusion MCRC algorithm is adopted to reconstruct the three-dimensional model of the segmentation results,and the model surface is optimized by smoothing.The results show that the method is of high practical value for it improves the ability of real-time interactive rendering,reduces the workload of manual annotation,effectively avoids the production of triangular patches,improves the display quality,and enhances the structural clarity and realism.