目的：检测磨牙非中性关系时虚拟咬合拼对的精度及临床可行性。方法：选取已完成正颌术前正畸、拼对终末咬合关系良好的寄存正颌患者牙列模型20副，其中10副术前正畸时上颌拔除左右各一颗双尖牙，术后拼对咬合为磨牙非中性关系（拔牙组）， 另10副未进行减数拔牙，术后拼对咬合为中性磨牙关系（非拔牙组）。手工拼对石膏模型确立的终末咬合关系作为评价拔牙组、非拔牙组误差的实测值。进行虚拟终末咬合关系确立的数字流程为：结构光扫描仪分别获取上下牙列的三维数据，导入 3Shape软件，在虚拟环境下整体移动上颌牙列确立终末咬合关系。得到的测试上颌模型和手工拼对上颌模型以下颌模型为标准图形配准，采用直线距离法分析上颌牙列每颗牙齿的实验位置与手工拼对位置的差异，计算误差值。结果：拔牙组虚拟咬合拼对的全牙列平均误差为(0.60±0.36) mm，大于非拔牙组的(0.51±0.18) mm，差异有统计学意义（P<0.05）；拔牙组误差与1 mm相比差异无统计学意义（P>0.05），非拔牙组误差小于1 mm，差异有统计学意义（P<0.05）。结论：磨牙非中性关系虚拟咬合拼对的误差值高于磨牙中性关系虚拟咬合拼对的误差值，但其精度满足临床需求，具备应用价值。

Objective： To evaluate the accuracy of virtual occlusal definition in non-Angle class Ⅰ molar relationship, and to evaluate the clinical feasibility. Methods: Twenty pairs of models of orthognathic patients were included in this study. The inclusion criteria were: (1) finished with pre-surgical orthodontic treatment and (2) stable final occlusion. The exclusion criteria were: (1) existence of distorted teeth, (2) needs for segmentation, (3) defect of dentition except for orthodontic extraction ones, and (4) existence of tooth space. The tooth-extracted test group included 10 models with two premolars extracted during preoperative orthodontic treatment. Their molar relationships were not Angle class Ⅰ relationship. The non-tooth-extracted test group included another 10 models without teeth extracted, therefore their molar relationships were Angle class Ⅰ. To define the final occlusion in virtual environment, two steps were included: (1) The morphology data of upper and lower dentition were digitalized by surface scanner (Smart Optics/Activity 102; Model-Tray GmbH, Hamburg, Germany); (2) the virtual relationships were defined using 3Shape software. The control standard of final occlusion was manually defined using gypsum models and then digitalized by surface scanner. The final occlusion of test group and control standard were overlapped according to lower dentition morphology. Errors were evaluated by calculating the distance between the corresponding reference points of testing group and control standard locations. Results: The overall errors for upper dentition between test group and control standard location were (0.51±0.18) mm in non-tooth-extracted test group and (0.60±0.36) mm in tooth-extracted test group. The errors were significantly different between these two test groups (P<0.05). However, in both test groups, the errors of each tooth in a single dentition does not differ from one another. There was no significant difference between errors in tooth-extracted test group and 1 mm (P>0.05); and the accuracy of non-tooth-extracted group was significantly smaller than 1 mm (P<0.05). Conclusion: The error of virtual occlusal definition of none class I molar relationship is higher than that of class I relationship, with an accuracy of 1 mm. However, its accuracy is still feasible for clinical application.