|
发布时间: 2018-01-16 |
医学图像处理 |
|
|
收稿日期: 2017-03-05; 修回日期: 2017-09-19
基金项目: 国家自然科学基金项目(51405197);浙江省自然科学基金项目(LQ14E050006)
第一作者简介:
吴婷(1979-), 女, 副教授, 2012年于南京航空航天大学获机械工程专业博士学位, 主要研究方向为数字化设计与制造, 生物医学工程。E-mail:wuting628@163.com.
中图法分类号: TP391
文献标识码: A
文章编号: 1006-8961(2018)01-0084-11
|
摘要
目的 从3维牙颌模型上分割出单颗牙齿是计算机辅助正畸系统的重要步骤。由于3维测量分辨率和网格重建精度的有限性,三角网格牙颌模型上牙龈和牙缝边界往往融合在一起,使得单颗牙齿的自动分割变得极为困难。传统方法容易导致分割线断裂、分支干扰等问题,且手工交互较多,为此提出一种新颖的基于路径规划技术的单颗牙齿自动分割方法。方法 为避免在探测边界时牙龈和牙缝相互干扰,采用牙龈路径和牙缝路径分开规划策略。首先基于离散曲率分析和一种双重路径规划法搜索牙龈分割路径,并基于搜索到的牙龈路径利用图像形态学和B样条拟合技术构建牙弓曲线;然后综合牙龈路径和牙弓曲线的形态特征探测牙龈路径上的牙缝凹点以划界每颗牙齿的牙龈边界轮廓,并通过匹配和搜索牙龈边界轮廓上颊舌侧凹点间的最优路径确定齿间牙缝边界路径;最后细化整个路径以获取每颗牙齿精确的封闭分割轮廓。结果 对不同畸形程度的患者牙颌模型进行分割实验,结果表明,本文方法对于严重畸形的牙齿能够产生正确的分割结果,而且简单快速,整个分割过程基本能够控制在20 s以内。和现有方法相比,本文方法具有较少的人工干预和参数调整,除了在个别牙齿边界较为模糊的位置需要手动调整外,大部分情况都是自动的。结论 提出的路径规划方法具有强大的抗干扰能力,能够有效克服牙缝牙沟等分支干扰以及分割线断裂等问题,最大程度地减少人工干预,适用于各类畸形牙患者模型的牙齿分割。
关键词
牙齿分割; 路径规划; 离散曲率; B样条拟合; 口腔正畸; 3维牙颌模型
Abstract
Objective Segmentation of individual tooth from 3D dental meshes is an important process in computer-aided orthodontic system. However, gingival margin and tooth interstices usually overlap or fuse due to limited measuring resolution and mesh-triangulating precision, which causes the automatic segmentation of teeth directly from 3D dental models into individual tooth objects to be extraordinarily difficult, especially when dealing with models with severe malocclusion problems. Traditional approaches easily lead to boundary breaks, branch interferences, and manual interactions. Therefore, a novel and automatic tooth segmentation method based on path planning is proposed in this study. As path-planning method can provide mathematically piecewise optimal boundaries while greatly reducing sensitivity to local noise and intervening structures, such as tooth interstices and fossae; it can reduce user interaction and repair time to a great extent. Method The proposed strategy avoids the limitations of previous methods by initially searching gingiva-teeth boundary paths and then tooth-tooth interstitial boundary paths to avoid interference between gingiva and interstice in the course of detecting boundaries. Dental differential characteristics, image morphology, and B-spline-fitting technology are also employed to ensure stability and accuracy. The feature region of interest between gingiva and teeth is firstly extracted using a discrete curvature analysis. The optimal gingiva-teeth boundary paths are detected based on this feature region by using a novel double-path-planning algorithm. The algorithm initially searches the gingiva-teeth paths on the basis of the vertex distance information of the feature region to ensure that the paths avoid the branching points at tooth interstices. Accurate gingiva-teeth paths are then searched by formulating the neighborhood set of initial gingiva-teeth paths, their vertex distance and curvature information to ensure that the paths adhere to the high-curvature locations between gingiva and teeth. Afterward, the searched gingiva-teeth paths are projected on the occlusal plane to form a gingiva-teeth path binary image, and the dental arch curve is automatically constructed using image morphology and a B-spline-extended-fitting technology. The interstitial concave corners on the gingiva-teeth paths are detected and deleted by combining the normal vectors and curvature characteristics of the gingiva-teeth boundary paths and the ones of the dental arch curve to demarcate the gingival boundary of each individual tooth. The tooth-tooth interstitial boundary paths are then obtained by searching the optimal paths from the endpoints on the lingual side of the gingival boundary of each tooth to their corresponding ones on the buccal side. The tooth-tooth interstitial paths and the gingiva-teeth paths constitute the closed boundary of each individual tooth. These path boundaries can be further refined by using a simple yet efficient method based on bipartition and path searching. Result Experiments on numerous dental models of patients with different levels of crowding problems are conducted to verify the efficiency and accuracy of the proposed method. The segmentation performance, time consumption, and user interaction of the proposed method are analyzed and compared with those of other published approaches. Results demonstrate that although some models include considerable noise and intervened branches, such as tooth fossae and grooves, the proposed method can easily remove these interference structures and produce good segmentation results even for severely deformed teeth and complex tooth arrangements. Besides, the proposed method involves fewer user interactions and parameter adjustments. And the procedures are automatic, except from setting the default curvature threshold for some models. In special cases where the tooth boundary is ambiguous, additional interactions are needed to repair undetected regions within this system interface manually. However, all these interactions are simple and time saving. The time consumption in manually repairing few missing or unwanted boundaries is generally less than 10 s. The key procedures (i.e., the gingiva-teeth path planning and the tooth-tooth interstitial path planning) only take less than 1 s, which greatly reduce the searching time compared with that of other methods. The most time-consuming operations are feature region extraction (5~6 s) and path refining (4~5 s). Along with the entire program execution time, the entire execution in each segmentation experiment can be finished in less than 20 s. Conclusion This study proposes a novel automated approach for segmenting individual tooth from dental meshes based on path planning. The approach utilizes the strong anti-interference and anti-fracture capabilities of path planning to avoid local noise, intervening structures, and boundary breaks. The proposed method can effectively overcome the interferences of interstice and fossa and the difficulty of path walking around tooth interstice branches by combining multiple-path planning, image morphology, and B-spline-fitting technology. Therefore, good results can be obtained even in the presence of distorted tooth shapes and complex tooth arrangements. Furthermore, the method is fast and effective and involves fewer user interactions and parameter adjustments compared with published approaches. The experimental results demonstrate that the proposed approach can address different levels of crowding problems. Therefore, this approach can be applied to clinical orthodontic planning treatments to improve their accuracy and efficiency. The limitation of the approach is that in cases where the convex and concave features of the gingival boundary is not distinct, the method may require user interactions to guarantee accurate results. However, all required interactions are simple and time saving. In the future, additional prior knowledge and artificial intelligence will be fused into the proposed framework to further enhance its efficiency and robustness.
Key words
tooth segmentation; path planning; discrete curvature; B-spline fitting; orthodontics; 3D dental meshes
0 引言
随着3维数字化成像技术在口腔医学领域的飞速发展,计算机辅助诊疗修复越来越多的应用在口腔修复当中,并逐渐成为这一领域的发展趋势[1]。在计算机口腔正畸系统当中,一个关键的步骤就是将数字化3维牙颌模型上的牙齿一个个单独分割出来,而分割的精度和效率直接关系到后续的虚拟牙排列和治疗规划的结果[2]。
受3维测量的分辨率和网格重建精度的限制,三角网格牙颌模型上龈缘和牙齿缝隙往往存在着粘连问题,再加上牙齿表面凹凸不平,使得单颗牙齿的自动分割变得极为困难,尤其当遇到患者存在牙齿畸形的状况,这一问题变得更为严重。尽管目前出现了许多商业化的牙科软件系统,例如“3shape”,实现了一定程度的牙齿分割自动化,但是它们对手工干预的程度较高,分割效率较低。由于牙齿形状各异,并且不同人的牙齿排列也有很大的差异,因此单颗牙齿的分割即使在牙科计算机软件中也是一项较为耗时的工作。
近几年来,国内外许多研究人员对单颗牙齿的自动分割进行了研究。这些方法主要分为两种类型:基于2维图像的分割方法和基于3维模型的分割方法。第一种类型最为典型的代表是Kondo等人[3]提出的深度图像分割法:首先从咬合平面方向上获取牙颌模型的深度图像,并通过计算该图像的牙弓曲线获得全景深度图像,然后结合两幅图像的边缘特征得到牙齿分割轮廓;然而,由于牙缝特征点是通过做牙弓曲线的垂线来获得,当患者存在畸形牙的时候,这种方法可能会产生较大的分割偏差。Grzegorzek等人[4]提出一种基于多层深度图像分割方法,避免了仅依靠单一图像而产生精度差的问题,该方法首先探测每一层图像的牙齿外轮廓和牙齿之间的非凸顶点,然后连接这些非凸顶点并映射回3维空间当中获得牙齿分割边界;然而这些深度图像也仅仅是从咬合平面视角方向获得,当患者存在不整齐的牙齿时,这种方法也不能产生较好的分割结果。
由于基于2维图像的分割无法准确表达牙齿的真实形态,因此基于3维模型的分割方法越来越受到研究学者的关注。Kronfeld等人[5]结合snake活动轮廓模型和边缘相切流的方法实现了牙齿的自动分割,然而由于相切流容易受到噪声的影响,因此分割精度较低。Yuan等人[6]首先删除齿间粘连区域,然后采用曲面能量约束方式恢复缺失的开口区域以获得单颗牙齿完整形态,最后利用骨架线方法获得每颗牙齿分割边界,但该方法在识别粘连区域时需要大量的手动交互。Kumar等人[7]首先利用曲率分析和漫水填充(flood fill)算法探测牙齿和牙龈的分割边界,然后探测此边界上的角点来获取牙缝分割边界;但漫水填充算法在噪音和干涉分支的干扰下容易产生欠分割或过分割的现象,该方法还需要复杂的路径搜索算法进行修复。Wu等人[8]提出了一种改进的形态学骨架分割算法较为精确的获取牙齿分割边界,但骨架线算法容易产生较多的骨刺和干涉分支。Zou等人[9]和Wang等人[10]提出一种基于调和场的牙齿分割方法,不仅能够处理严重畸形的牙齿,还能够处理牙齿边缘模糊的情况,然而该方法需要手动设定每颗牙齿的分割约束点,因此人工干预较多。
针对上述问题,本文提出一种新颖的利用路径规划技术的单颗牙齿自动分割方法,以利用路径搜索强大的抗干扰抗断裂能力最大程度地减少人工干预。该方法同时结合牙颌模型的微分特征、图像形态学以及B样条拟合技术,以确保分割的精度和稳定性。
1 技术路线
由于三角网格牙颌模型上牙齿和牙龈以及牙齿和牙齿之间的边界往往交织在一起,因此,为避免在探测分割边界时产生相互干扰,首先规划牙齿和牙龈之间的分割路径,然后对牙齿之间的牙缝分割路径进行规划,具体路线如下:
1) 牙龈分割路径规划(图 1)。首先利用模型的曲率特征提取牙齿和牙龈分界的特征区域,并基于一种双重路径规划法搜索牙龈分割路径,然后对得到的路径利用图像形态学和B样条延伸拟合方法构建牙弓曲线。
2) 牙缝分割路径规划(图 2)。综合牙龈路径和牙弓曲线的形态特征搜索牙龈路径上的凹点,以确定牙齿之间的缝隙边界点,然后基于路径规划法搜索牙缝分割路径,最后细化路径以得到每颗牙齿精确的分割轮廓。
2 牙龈分割路径规划
2.1 特征区域提取
牙齿和牙龈之间的分界特征区域一般位于牙颌模型表面曲率较大的地方,因此首先采用局部曲面拟合法[11]对模型进行离散曲率分析,并基于最大主曲率原则获取牙颌模型上每个顶点
对变换后的曲率值进行阈值操作{
2.2 牙龈路径搜索
路径规划方法是一种基于图论的节点路径搜索算法,其最大的特点是在获取最优分割线的同时,避免分割线断裂和干涉分支。但由于其搜索方向无法控制,直接进行路径搜索,容易使路径从牙缝处绕行,导致不正确的结果,如图 4所示。因此,采用一种基于双重规划的牙龈路径搜索方法,以确保分割结果的稳定性。
2.2.1 确定起点和终点
路径规划需要首先确定路径的起点和终点,为减少人工干预,利用坐标转换法进行自动搜索。将特征区域
2.2.2 搜索牙龈路径
路径搜索的基本方法是将数据构成一个带权的图,图中两个节点之间边的权值表示相应的代价,最优路径就是从起点到终点中所有边的代价之和最小的那条路径[14]。根据特征区域
$ {w_1}\left( {{\mathit{\boldsymbol{v}}_i}, {\mathit{\boldsymbol{v}}_j}} \right) = \left\{ \begin{array}{l} \left\| {{\mathit{\boldsymbol{v}}_i} - {\mathit{\boldsymbol{v}}_j}} \right\|\;\;\;\;\;\left( {{\mathit{\boldsymbol{v}}_i}, {\mathit{\boldsymbol{v}}_j}} \right) \in \mathit{\boldsymbol{E}}\\ \infty \;\;\;\;\;\;\;\;\;\;\;\;\;其他 \end{array} \right. $ | (1) |
式中,
值的注意的是,舌侧牙龈路径搜索后,需更新路径的起点和终点,如图 5(c)所示,因为利用坐标转换法得到的起点和终点可能会偏离所需的位置。
2.2.3 修正牙龈路径
由于第1次搜索路径为避开牙缝,得到的路径仅仅是“最短的”,并没有很好地贴合在牙齿和牙龈分界最为凹陷的地方,因此利用模型的离散曲率信息对路径进行修复。首先收集上一步搜索得到的舌侧牙龈路径
$ {w_2}\left( {{\mathit{\boldsymbol{v}}_i}, {\mathit{\boldsymbol{v}}_j}} \right) = \left\{ \begin{array}{l} \frac{{\left\| {{\mathit{\boldsymbol{v}}_i} - {\mathit{\boldsymbol{v}}_j}} \right\|}}{{\kappa \left( {{\mathit{\boldsymbol{v}}_\mathit{i}}} \right)\kappa \left( {{\mathit{\boldsymbol{v}}_\mathit{j}}} \right)}}\;\;\;\;\;\left( {{\mathit{\boldsymbol{v}}_i}, {\mathit{\boldsymbol{v}}_j}} \right) \in \mathit{\boldsymbol{E}}\\ \infty \;\;\;\;\;\;\;\;\;\;\;\;\;其他 \end{array} \right. $ | (2) |
式中,
2.3 牙弓曲线拟合
牙弓曲线是医学上描述牙齿在牙槽骨上排列的形态。在口腔正畸中,牙弓曲线是评定牙齿咬合不正的一个重要参考标准。目前,用来描述牙弓曲线的几种数学形式,主要有多项式曲线,悬链曲线、beta函数曲线以及B样条曲线等,由于在实际的操作过程中,医生可能会对患者的牙弓曲线进行修改[16],因此利用B样条曲线的局部修改性质进行调整比较方便。但传统方法进行B样条曲线拟合时,往往需要人工交互选择特征型值点,本文提出一种基于图像形态学和B样条延伸拟合的方法对搜索到的牙龈分割路径自动创建牙弓曲线。
首先将搜索到的牙龈分割路径投影到咬合平面上,构建一个牙龈路径二值图像(图 6(a)),然后对该二值图像进行形态学填充(图 6(b))以及细化操作,以获取牙弓骨架线(图 6(c))。
将牙弓骨架线作为型值点数据
$ L = \sum\limits_{i = 2}^m {\left\| {{\mathit{\boldsymbol{Q}}_i} - {\mathit{\boldsymbol{Q}}_{i - 1}}} \right\|} $ | (3) |
则延伸后的总长可表示为
$ {L_{{\rm{extend}}}} = L\left( {1 + {\alpha _0} + {\alpha _m}} \right) $ | (4) |
因此,B样条曲线
$ \left\{ \begin{array}{l} {s_1} = \frac{{{\alpha _0}}}{{1 + {\alpha _0} + {\alpha _m}}}\\ {s_i} = {s_{i - 1}} + \frac{{\left\| {{\mathit{\boldsymbol{Q}}_i} - {\mathit{\boldsymbol{Q}}_{i - 1}}} \right\|}}{{{L_{{\rm{extend}}}}}}\;\;\;\;\;\;i = 2, 3, \cdots, m \end{array} \right. $ | (5) |
通过最小化能量函数
$ E = \sum\limits_{i = 1}^m {\left\| {\mathit{\boldsymbol{C}}\left( {{s_i}} \right) - {\mathit{\boldsymbol{Q}}_i}^2} \right\|} + \lambda \sum\limits_{i = 1}^m {{{\left\| {\mathit{\boldsymbol{C''}}\left( {{s_i}} \right)} \right\|}^2}} \Delta s $ | (6) |
即可获得延伸拟合后的牙弓B样条曲线
3 牙缝分割路径规划
3.1 凹点探测
牙齿之间的缝隙位置一般位于牙龈路径的凹点处,如图 7所示,因此在牙缝路径规划前,需先进行凹点探测。由于牙龈路径波浪起伏不定,这里提出一种综合牙弓曲线和牙龈路径形态特征的自动探测方法。
曲线上的凹点和凸点一般都具有较大的曲率,因此首先将得到的舌侧牙龈路径
仔细观察牙龈曲线上的凹点特性,可以发现,凹点的法向量方向与凹点距离牙弓曲线上的最近点和该凹点连线的方向一致(图 8(c)(d)),而法向量可以利用二阶导数确定,因此可利用约束条件进行凹点探测,即
$ \begin{array}{l} \mathit{\boldsymbol{f}}_{{\rm{in}}}\left( t \right) = \left\{ \begin{array}{l} 凹点\;\;\;\;\;\;\;\;{{\mathit{\boldsymbol{f''}}}_{{\rm{in}}}}\left( t \right) \cdot \left( {{\mathit{\boldsymbol{f}}_{{\rm{in}}}}\left( t \right) - \mathit{\boldsymbol{C}}\left( s \right)} \right) > 0\;\& \\ \;\;\;\;\;\;\;\;\;\;\;\;\;\;{\rho _{{\rm{in}}}}\left( t \right) \ge T\\ 非凹点\;\;\;\;\;\;\;\;\;\;其他 \end{array} \right.\\ \;\;\;\;\;\;\;\;\;\;\;\;\;\;\;\;\;\;\;\;\;\;\;{\mathit{\boldsymbol{f}}_{{\rm{out}}}}\left( t \right) = \\ \left\{ \begin{array}{l} 凹点\;\;\;\;\;\;\;\;{{\mathit{\boldsymbol{f''}}}_{{\rm{out}}}}\left( t \right) \cdot \left( {{\mathit{\boldsymbol{f}}_{{\rm{out}}}}\left( t \right) - \mathit{\boldsymbol{C}}\left( s \right)} \right) > 0\;\& \\ \;\;\;\;\;\;\;\;\;\;\;\;\;\;{\rho _{{\rm{out}}}}\left( t \right) \ge T\\ 非凹点\;\;\;\;\;\;\;\;\;\;其他 \end{array} \right. \end{array} $ | (7) |
式中,
将探测到的凹点(图 8(e))删除,并将剩余的舌侧和颊侧牙龈路径上的点按顺序进行组合,获得每颗牙齿的牙龈分割线,如图 8(f)所示。
3.2 牙缝路径搜索
由于牙缝位于模型表面相邻牙齿之间的凹状区域,因此也利用路径规划方法进行搜索。将特征区域
4 分割路径细化
利用路径规划方法得到的路径点都只是网格模型上的顶点,其路径的光滑性取决于网格的密度。因此,本文系统提供给用户一个额外的选择:细化路径。目前,关于细化分割线的方法大都采用插值[8]或网格细分[18]等方法,然而网格细分容易导致分割轮廓偏离原始路径,而插值法需要将插值点投影到网格上才能获得正确的结果, 且投影求交繁琐。本文采用一种的简单而有效的策略[19]以对获得的原始分割路径进行细化。
对于每颗牙齿封闭的路径,首先搜索其邻域网格,将路径邻域内的每一条网格特征边进行对分,使对分后的每段边长小于给定的值
$ \begin{array}{l} \left\{ \begin{array}{l} {\mathit{\boldsymbol{v}}_x} = {\mathit{\boldsymbol{v}}_i} + x\frac{{\left\| {{\mathit{\boldsymbol{v}}_i} - {\mathit{\boldsymbol{v}}_j}} \right\|}}{{N + 1}}\\ \kappa \left( {{\mathit{\boldsymbol{v}}_x}} \right) = \kappa \left( {{\mathit{\boldsymbol{v}}_i}} \right) + x\frac{{\left| {\kappa \left( {{\mathit{\boldsymbol{v}}_i}} \right) - \kappa \left( {{\mathit{\boldsymbol{v}}_j}} \right)} \right|}}{{N + 1}} \end{array} \right.\\ \;\;\;\;\;\;\;\;\;\;\;\;\;\;x = 1, 2, \cdots, N\\ \;\;\;\;\;\;\;\;\;\;\;N = floor\left( {\frac{{{\mathit{\boldsymbol{v}}_i} - {\mathit{\boldsymbol{v}}_j}}}{{{d_{{\rm{min}}}}}}} \right) \end{array} $ | (8) |
式中,
将细分后的所有顶点构建无向连通图
$ {w_3}\left( {{\mathit{\boldsymbol{v}}_x}, {\mathit{\boldsymbol{v}}_y}} \right) = \left\{ \begin{array}{l} \frac{{\left\| {{\mathit{\boldsymbol{v}}_x} - {\mathit{\boldsymbol{v}}_y}} \right\|}}{{\kappa \left( {{\mathit{\boldsymbol{v}}_x}} \right)\kappa \left( {{\mathit{\boldsymbol{v}}_y}} \right)}}\;\;\;{\mathit{\boldsymbol{v}}_x}, {\mathit{\boldsymbol{v}}_y}\ 属于同一三角片\\ \infty \;\;\;\;\;\;\;\;\;\;\;\;\;\;\;\;\;\;其他 \end{array} \right. $ | (9) |
然后,在此连通图内,基于每颗牙齿的牙齿分割轮廓,再次利用Dijkstra算法搜索精确的分割路径。图 10显示路径细化前后的对比结果,可以看出,细化后的路径更加光滑,同时也很好地贴合在边界较明显的位置。
5 实验分析
5.1 实验条件
5.2 结果分析
由于篇幅所限,图 11列出了实验中5种典型的牙颌模型的分割结果,并按行显示了具体的分割过程。这5个模型的牙齿畸形程度依次递增。可以看出,尽管在特征提取阶段,一些模型的特征区域包含很多牙沟牙窝以及噪音等干涉区域,但是在牙龈路径搜索过程中,仍然获得了较好的结果,这是因为路径规划具有强大的抗干扰能力,能够避免产生干涉分支;同时由于我们提出的双重路径规划方法,较好的避免了路径的绕行问题。另外,也可以发现,即使在像模型4和模型5这样牙齿畸形较严重的情况下,本文方法也能产生正确的结果。
在整个实验中,大部分过程基本都是自动的,需要调整的参数较少,除了一些模型需要修改默认的曲率阈值。但是,当存在牙齿分割边界较为模糊的情况下,例如像如图 12显示的模型,由于牙齿错位造成相邻牙齿分界位置较为不明显,因此在牙缝探测过程中可能会出现图 12中所示的遗漏现象,需要手工干预调整。但相比于3shape软件以及其他文献(如文献[9]设置每颗牙齿周围的约束点)的方法,本文的手工调整方便而简单,能够同时保证分割结果的效率和准确性。
5.3 耗时分析
为方便比较算法消耗时间,将整个系统的操作过程分为4个部分:“特征区域提取”、“牙龈分割路径规划”、“牙缝分割路径规划”和“分割路径细化”。各部分耗时如表 1所示。其中,“特征区域提取”包含离散曲率分析、阈值提取和连通性过滤,“牙龈分割路径规划”包含牙龈路径探测和牙弓曲线拟合,“牙缝分割路径规划”则包含凹点探测和牙缝路径搜索,最后的“分割路径细化”包含网格对分和牙齿切割。
/s | |||||
模型 (顶点数/三角片数) |
特征区域提取 | 牙龈分割路径规划 | 牙缝分割路径规划 | 分割路径细化 | 总时间 |
1 (170 756/339 847) | 5.35 | 0.72 | 0.63 | 4.22 | 10.92 |
2 (208 685/415 653) | 6.53 | 0.91 | 0.75 | 5.64 | 13.83 |
3 (219 418/436 930) | 6.88 | 0.84 | 0.73 | 4.32 | 12.77 |
4 (199 889/397 911) | 6.37 | 0.88 | 0.76 | 5.35 | 13.36 |
5 (201 092/400 484) | 6.56 | 0.93 | 0.74 | 4.81 | 13.04 |
从表 1可以看出,牙龈分割路径规划和牙缝分割路径规划都各自只需不到1 s的时间即可完成,相比于其他文献方法,大大降低了搜索时间。本文算法在特征区域提取和分割路径细化上耗时较多,其中特征区域提取平均需耗时约为5~6 s,并且其时间随网格模型顶点数量的增多而增加,该过程需要对网格的每个顶点进行局部曲面拟合以计算曲率,因此耗时较长;其次,分割路径细化耗时大约为4~5 s,这主要是因为细化路径需要花费较多的时间对原始分割路径周围的网格进行对分,并判断这些细分顶点之间是否属于同一个三角片,因此耗时较多。每个模型在整个分割过程中,加上打开模型和设置参数,基本控制在20 s以内。即使特殊情况加上手动干预,本系统也能在30 s之内完成一个牙齿模型的分割。而文献[8-9]都至少需要1~2 min的时间;著名的“3shape”软件则更是需要数分钟时间的手动交互才能完成分割。
6 结论
本文针对3维牙颌解剖形态和排列特点,提出一种新颖的基于路径规划技术的单颗牙齿自动分割方法。该方法采用牙龈分割路径和牙缝分割路径分开搜索的策略,避免了牙龈和牙缝分割线的相互干扰;提出的双重路径规划法避免了传统分割路径容易产生的分割线绕行以及断裂问题,并有效解决牙沟和牙缝干涉分支这一难点;结合图像形态学和B样条拟合的单颗牙齿路径划分方法,避免了传统方法导致误分以及人工干预较多的问题。与现有的方法相比,本文方法简单快速,而且具有较少的人工干预和参数调整,适用于各种畸形牙患者的牙齿分割,对后续正畸治疗中的虚拟牙排列具有重要应用价值。但对于牙齿边界较为模糊的情况,本方法还需要一些手动调整,如何进一步地减少手动干预、提高分割的精度,将是后续工作的重点。
参考文献
-
[1] Wu T, Zhang L B. Three dimention tooth reconstruction using level set active contour model[J]. Journal of Image and Graphics, 2016, 21(8): 1078–1087. [吴婷, 张礼兵. 水平集活动轮廓模型的3维牙齿重建[J]. 中国图象图形学报, 2016, 21(8): 1078–1087. ] [DOI:10.11834/jig.20160812]
-
[2] Kumar Y, Janardan R, Larson B. Automatic virtual alignment of dental arches in orthodontics[J]. Computer-Aided Design and Applications, 2013, 10(3): 371–398. [DOI:10.3722/cadaps.2013.371-398]
-
[3] Kondo T, Ong S H, Foong K W C. Tooth segmentation of dental study models using range images[J]. IEEE Transactions on Medical Imaging, 2004, 23(3): 350–362. [DOI:10.1109/TMI.2004.824235]
-
[4] Grzegorzek M, Trierscheid M, Papoutsis D, et al. A Multi-stage approach for 3D teeth segmentation from dentition surfaces[C]//Proceedings of the 4th International Conference on Image and Signal Processing. Berlin, Heidelberg:Springer, 2010, 6134:521-530.[DOI:10.1007/978-3-642-13681-8_61]
-
[5] Kronfeld T, Brunner D, Brunnett G. Snake-based segmentation of teeth from virtual dental casts[J]. Computer-Aided Design and Applications, 2010, 7(2): 221–233. [DOI:10.3722/cadaps.2010.221-233]
-
[6] Yuan T R, Liao W H, Dai N, et al. Single-tooth modeling for 3D dental model[J]. International Journal of Biomedical Imaging, 2010, 2010: #535329. [DOI:10.1155/2010/535329]
-
[7] Kumar Y, Janardan R, Larson B, et al. Improved segmentation of teeth in dental models[J]. Computer-Aided Design and Applications, 2011, 8(2): 211–224. [DOI:10.3722/cadaps.2011.211-224]
-
[8] Wu K, Chen L, Li J, et al. Tooth segmentation on dental meshes using morphologic skeleton[J]. Computers & Graphics, 2014, 38: 199–211. [DOI:10.1016/j.cag.2013.10.028]
-
[9] Zou B J, Liu S J, Liao S H, et al. Interactive tooth partition of dental mesh base on tooth-target harmonic field[J]. Computers in Biology and Medicine, 2015, 56: 132–144. [DOI:10.1016/j.compbiomed.2014.10.013]
-
[10] Li Z Y, Wang H. Interactive Tooth separation from dental model using segmentation field[J]. PLoS One, 2016, 11(8): e0161159. [DOI:10.1371/journal.pone.0161159]
-
[11] Milroy M J, Bradley C, Vickers G W. Segmentation of a wrap-around model using an active contour[J]. Computer-Aided Design, 1997, 29(4): 299–320. [DOI:10.1016/S0010-4485(96)00058-9]
-
[12] Wu T, Zhang L B. Curvature analysis and visualization of dental mesh models[C]//Proceedings of the 2016 Joint International Conference on Artificial Intelligence and Computer Engineering and International Conference on Network and Communication Security. Lancaster, U.S.A:DEStech, 2016:77-83.[DOI:10.12783/dtcse/aice-ncs2016/5618]
-
[13] Rössl C, Kobbelt L, Seidel H P. Extraction of feature lines on triangulated surfaces using morphological operators[C]//AAAI Spring Symposium on Smart Graphics. Menlo Park, USA:AAAI Press, 2000:71-75.
-
[14] Gallo G, Pallottino S. Shortest path algorithms[J]. Annals of Operations Research, 1988, 13(1): 1–79. [DOI:10.1007/BF02288320]
-
[15] Dijkstra E W. A note on two problems in connexion with graphs[J]. Numerische Mathematik, 1959, 1(1): 269–271. [DOI:10.1007/BF01386390]
-
[16] Kumar Y, Janardan R, Larson B E. Automatic feature identification in dental meshes[J]. Computer-Aided Design and Applications, 2012, 9(6): 747–769. [DOI:10.3722/cadaps.2012.747-769]
-
[17] Wu T, Zhang L B. Auto-detection of arch curve from dental meshes[C]//Proceedings of the 4th International Conference on Machinery, Materials and Information Technology Applications. Paris, France:Atlantis, 2016, 71:1443-1447.[DOI:10.2991/icmmita-16.2016.265]
-
[18] Lee Y, Lee S, Shamir A, et al. Intelligent mesh scissoring using 3D snakes[C]//Proceedings of the 12th Pacific Conference on Computer Graphics and Applications. Washington DC, USA:IEEE, 2004:279-287.[DOI:10.1109/PCCGA.2004.1348358]
-
[19] Le B H, Deng Z G, Xia J, et al. An interactive geometric technique for upper and lower teeth segmentation[C]//Proceedings of the 12th International Conference on Medical Image Computing and Computer-Assisted Intervention. Berlin, Heidelberg:Springer, 2009, 5762:968-975.[DOI:10.1007/978-3-642-04271-3_117]