论文部分内容阅读
目的：在动物骨上模拟真实的手术环境，探讨不同取点数目对基于术前CT导航的股骨配准精度影响，确定真实临床手术中最少的取点数目。方法：在包含完整皮肤、肌肉的猪腿骨上进行不同取点数目的配准试验，通过Matlab软件对点云数据进行处理，再经Geomagic软件进行封装及配准，测试目标点的配准误差（targetregistrationerror，TRE），最后通过SPSS软件对相关参数进行统计学分析。结果：取点数目越多，TRE越小；每片点云取15个点，配准结果满意，此时股骨近端点云集配准误差为（0.804±0.425）mm，股骨远端点云集配准误差为（0.821±0.448）mm；当每片点云取点数目大于等于15时，各组与每片点云25个点间相比TRE无统计学差异。结论：取点数量越多配准精度越高，结合实际临床情况，每片点云应最少取15个点。“,”Objective: To investigate the influence of different number of points involved to the accuracy of femora registration by simulating real operating condition using animal bone and to ascertain the minimum number of points involved in real clinical operation. Methods: Registration experiments based on various number of points involved were carried out on pig femur which contained complete skin and muscle. Matlab was applied here for point cloud data processing. Data was then imported into Geomagic Studio software to analyze and register. Target registration error (TRE) was also measured. Finally, SPSS was used to analyze the collected data. Results: The more points involved in the femur registration we got the smaller TRE. When the number of each point cloud was 15, the registration was fine, in another word, the error of registration of proximal femoral point cloud was limited to (0.804±0.425)mm, besides, the error of registration of distal femoral point cloud was limited to (0.821±0.448)mm. When the number of point cloud was larger or equals to 15 in these groups, TRE showed no statistic difference comparing with results when the number of points involved of each point was 25. Conclusion: The more points involved in registration the higher accuracy could be achieved. Considering clinical practice, the number of points involved of each point cloud should be larger or equals to 15.