摘要
目的:比较迭代最近点法(interactive closet point,ICP)、普氏分析法(Procrustes analysis,PA)和点构法在确定大视野锥形束CT(cone beam computed tomography,CBCT)数据正中矢状面(mid-sagittal plane,MSP)的准确性,为大视野锥形束CT空间坐标系的建立和对称性分析提供理论依据。方法:10名面部基本对称正畸患者进行大视野锥形束CT扫描,数据以DICOM格式保存输出,在Mimics10.0下完成数据分割获取数字化三维头颅,将所生成数字化头颅数据导入逆向工程软件geomagic studio 2012,对原始数据进行左右镜像,将原始和镜像后的三维头颅数据分别以ICP和PA进行配准,分别求得颅面结构正中矢状参考平面S1、S2。点构法是在In Vivo Dental 5.0软件下以蝶鞍点(sella,S)、鼻根点(nasion,N)、颅底点(basion,Ba)3点确定正中矢状参考平面S3,分别测量计算9对对称解剖标志点到上述3个参考平面的距离差值的绝对值,使用单因素方差分析比较3种方法生成正中矢状面的差异,并以LSD进行组间两两比较。结果:在正位截图下观察,3种方法生成的正中矢状参考平面均可用于临床分析,单因素方差分析显示3种生成正中矢状参考平面的方法差异有统计学意义(F=10.932,P=0.001),LSD检验显示迭代最近点算法与点构法生成MSP差异无统计学意义(P=0.11),普氏分析算法与点构法生成MSP差异有统计学意义(P=0.01)。结论:对于大视野锥形束CT数据,基于迭代最近点算法和普式分析算法计算可构建三维头颅的正中矢状参考平面,对于面部基本对称个体,迭代最近点算法与传统点构法所生成正中矢状参考平面无明显差异,具有临床可行性。
Objective： To compare the accuracyof interactive closet point（ ICP） algorithm,Procrustes analysis（ PA） algorithm,andalandmark-independent method to construct the mid-sagittal plane（ MSP） of the cone beam computed tomography. To provide theoretical basis for establishing coordinate systemof CBCT images and symmetric analysis. Methods： Ten patients were selected and scanned by CBCT before orthodontic treatment. The scan data was imported into Mimics 10. 0 to reconstructthree dimensional skulls. And the MSP of each skull was generated by ICP algorithm,PA algorithm and landmark-independent method. MSP extracted by ICP algorithm or PA algorithm involvedthree steps. First,the 3D skull processing was performed by reverse engineering software geomagic studio 2012 to obtain the mirror skull.Then,the original and its mirror skull was registered separately by ICP algorithm in geomagic studio 2012 and PA algorithm in NX Imageware 11. 0. Finally,the registered data were united into new data to calculate the MSP of the originaldata in geomagic studio 2012. The mid-sagittal plane was determined by SELLA（ S）,nasion（ N）,basion（ Ba） as traditional landmark-dependent methodconducted in software In Vivo Dental 5. 0. The distance from 9 pairs of symmetric anatomical marked points to three sagittal plane were measured and calculated to compare the differences of the absolute value. The one-way ANOVA test was used to analyze the variable differences among the 3 MSPs. The pairwise comparison was performed with LSD method. Results： MSPs calculated by the three methods were available for clinic analysis,which could be concluded from the front view. However,there was significant differences among the distances from the 9 pairs of symmetric anatomical marked points to the MSPs（ F = 10. 932,P =0. 001）. LSD test showed there was no significant difference between the ICP algorithm and landmarkindependent method（ P = 0. 11）,while there was significant difference between the PA algorithm and landmark-independent methods（ P = 0. 01）. Conclusion： Mid-sagittal plane of 3D skulls could be generated base on ICP algorithm or PA algorithm. There was no significant difference between the ICP algorithm and landmark-independent method. For the subjects with no evident asymmetry,ICP algorithm is feasible in clinical analysis.
出处
《北京大学学报（医学版）》
CAS
CSCD
北大核心
2016年第2期330-335,共6页
Journal of Peking University:Health Sciences
基金
国家高技术研究发展计划(863计划
2013AA040803)项目资助~~
