对基于标志点的颅面部三维影像重叠方法的改进及新的上颌骨三维影像重叠方法  

Optimization for landmarks-based three-dimensional cephalometric superimposition and a method for superimposition of three-dimensional image of maxilla

作　　者：翁瑶[1] 王紫仪[1] 张晓东[1] 

机构地区：[1]沈阳军区总医院口腔内科,辽宁沈阳110840

出　　处：《中国医学影像技术》2017年第1期124-131,共8页Chinese Journal of Medical Imaging Technology

基　　金：辽宁省自然科学基金(2015020428)

摘　　要：目的优化现有的基于标志点的三维影像重叠方法,并在此基础上提出一种更为简便的用以重叠上颌骨三维影像方法。方法对9名志愿者均间隔1年拍摄2幅锥形束CT(CBCT)影像,比较两种数学运算方法(6点优化法和8点优化法)对建立ELSA共坐标系误差的优化效果。以前后鼻棘点(ANS、PNS)建立用以重叠上颌骨的共坐标系(ANS、PNS共坐标系)。挑选数个标志点检测影像的重叠误差,并进行统计学分析。结果在未优化、6点优化、8点优化3种方法中,ELSA共坐标系建立的平均线距和角度误差分别为(0.60±0.47)mm和(0.49±0.38)°、(0.44±0.39)mm和(0.48±0.37)°、(0.41±0.34)mm和(0.41±0.32)°,8点优化法对ELSA共坐标系建立的平均线距误差小于无优化和6点优化法(P<0.05)°ELSA共坐标系下各标志点重叠的平均线距和角度误差分别为(0.95±0.50)mm和(1.06±0.65)°、(0.92±0.54)mm和(0.92±0.58)°、(0.80±0.43)mm和(0.87±0.59)°,3种方法差异有统计学意义(P均<0.000 1)。经8点优化后,在ELSA、ANS、PNS共坐标系中用以重叠上颌骨的各标志点的平均线距和角度误差分别为(0.92±0.47)mm和(1.14±0.66)°、(0.99±0.47)mm和(1.33±0.64)°、(1.00±0.62)mm和(1.08±0.70)°。结论本研究提出的8点优化法可显著提高ELSA共坐标系的准确性,且所提出的上颌骨三维影像重叠方法的误差已基本达到进行临床病例分析的要求范围。Objective To optimize the existed landmarks-basecrsupenmposition technique,and to propose a new simple method for regional superimposition of the maxillary complex.Methods Totally 9 volunteers were included,who were performed twice cone beam CT（CBCT） in one-year interval.ELSA co-ordinate were corrected using two kind of numerical optimization algorithm（6-point method and 8-point method）,and the operator error of landmark location were compared.The anterior and posterior nasal spine（ANS and PNS） co-ordinate system were constructed only using two anatomical landmarks（ANS and PNS） for regional superimposition of the maxillary complex.Several landmarks were pick up to evaluate the error of superimposition and analyze the difference by statistic method.Results In correction,6-point method and8-point method,the average error of linear and angle measurement of the ELSA co-ordinate were（0.60 ± 0.47） mm and（0.49 ± 0.38）°,（0.44 ± 0.39）mm and（0.48 ± 0.37）°,（0.41 ± 0.34）mm and（0.41 ± 0.32）° respectively.The average error of linear measurements significantly decreased（P〈0.05） in 8-point method.The average error of superimposition of all picked landmarks on ELSA co-ordinate were（0.95 ± 0.50）mm and（1.06 ± 0.65）°,（0.92 ± 0.54）mm and（0.92 ±0.58）°,（0.80 ± 0.43）mm and（0.87 ± 0.59）° respectively.The differences had statistical significance（P〈0.000 1）.In optimized by 8-point method,the average error of superimposition of all picked landmarks on ELSA,ANS and PNS co-ordinate were（0.92 ± 0.47）mm and（1.14 ± 0.66）°,（0.99 ± 0.47）mm and（1.33 ± 0.64）°,（1.00 ± 0.62）mm and（1.08 ±0.70）° respectively.Conclusion Significant improvement of superimposition is observed using 8-point method.The improved method is accuracy to analyze the clinical data.

关 键 词：锥形束计算机体层摄影术 三维头影重叠 优化 算法 头影测量学 

分 类 号：R814.42[医药卫生—影像医学与核医学]