Objective To analyze the distributed characteristics and producing reason of point-to-point registration errors within the registration area by repeated experiments. Methods Point-to-point registration model was set up by a femoral sawbone and Stryker' s computer navigation system. Registration points and standardization points were made in the surface area of distal femoral sawbone. DICOM data were imported to navigation software after CT scanning of sawbone. Three registration points and 1 standardization point were chosen at random to make point-to-point registrations for 48 times repeat-ly. The errors calculated by navigation software were restricted below 1 mm and all the coodinates of 192 registration points were recorded before and after registration. Actual 3D errors could be obtained then. The actual errors of registration points should be analyzed statistically for their distributed characteristics and their difference and correlation with software errors. Result The distribution of actual errors showed skewness of all 192 registration points. There was no correlation statistically between software errors and actual errors (P =0.904). Meanwhile, the difference between these two kinds of errors had significance ( P = 0.000 ). Conclusions There would be very high accuracy and accurate rate in the registration area after point-to-point registration. The errors calculated by navigation software could not represent the actual errors, nevertheless it would not influence actual accuracy of navigation system.%目的 通过点对点注册实验的方法来分析注册区域内的误差分布状况和产生的原因.方法 采用Stryker 公司的计算机导航系统将股骨模型建立为计算机导航点对点注册模型,于股骨模型远端和干骺端制作38个注册点和4个标定点,行薄层CT扫描后将DICOM数据导入计算机导航软件中.随机选取3个注册点和1个标定点作为点对点注册的4个注册点,应用重复进行点对点注册的方法进行了48组实验.限定注册后192个注册点的软件误差小于1 mm并记录其数值,记录注册前后192个注册点的坐标值来计算实际误差,并对注册点实际误差的分布状况以及软件误差和实际误差的差异和关系进行统计学分析.结果 192个注册点的实际误差分布呈偏态分布,软件误差和实际误差不具有统计学相关性(P =0.904),而二者的差异有统计学意义(P=0.000).结论 点对点注册区域内的准确性和准确率均很高,导航系统计算的误差并不能代表实际误差,但并不影响导航系统的实际准确性.
展开▼
