A novel segmentation method for medical image with intensity inhomogeneity is introduced.In the proposed active contour model,both region and gradient information are taken into consideration.The former,i.e.,region-based fitting energy,draws upon the region information and guarantees the accurate extraction of inhomogeneous image's local information.The latter,i.e.,an edge indicator,weights the length penalizing term to consider the gradient constrain.Moreover,signed distance penalizing term is also added to ensure accurate computation and avoid the time-consuming re-initialization of evolving level set function.Experiments for synthetic and real images demonstrate the feasibility and superiority of the proposed model.
目的:探索建立基于虚拟现实技术的正颌手术模拟研究方法。方法:将颌面畸形患者CT断层图像利用Simplant软件行颅颌面骨组织三维重建,截骨,构建颅颌面骨树状结构模型。在虚拟环境碰撞模型上行上颌骨Le Fort Ⅰ型截骨、双侧下颌支矢状劈开及颏成形的手术模拟。结果:该系统形成的三维立体虚拟影像,可从任意视角观察,清晰逼真。运用系统工具进行骨组织的旋转、平移等手术模拟操作,实现了操作者与模型的交互作用,图像及触觉感知实时反馈,沉浸感强。应用该方法实现了正颌手术设计、虚拟操作及术后效果预测。结论:颅面三维虚拟正颌手术系统可以行正颌手术模拟及手术操作训练,在虚拟环境下实现图像反馈和力、触觉感知的手术模拟,具有较高的临床应用价值。
