目的：本研究的目的是获取中国男性健康膝关节的解剖参数范围，并验证基于健康膝关节所建立的统计学形状模型(Statistical Shape Model, SSM)在几何外形上的有效性，为基于膝关节SSM的假体设计提供依据。 方法：采集健康男性膝关节CT影像(n>100)建立三维膝关节模型，每一例模型都将被选取为目标样本，剩余模型作为训练集进行主成分分析(Principal Component Analysis, PCA)，将得到的膝关节SSM向着目标样本模型进行拟合预测得到SSM预测模型。在样本模型和SSM预测模型上分别标记出相同的解剖测量点(股骨上6个，髌骨上8个，胫骨上6个)，并得出20个解剖参数(17个线性参数和3个角度参数)。对解剖参数值采用独立样本t检验(参数验证)和Mann-Whitney U检验(非参数验证)进行统计学检验，若得出的p-value均大于0.05，则证明SSM在几何外形上的有效性。 结果：PCA得到的主成分vi表示训练集模型中不同的解剖变异性，如主成分1表示训练集模型中体积差异，表现为膝关节骨成分整体的放大和缩小，主成分2和3分别表示尺寸上的不同比例关系和膝关节一定的屈膝旋转变化。本研究得出了中国男性健康右膝关节的解剖参数范围，如LAP、TEA和∠MA的参考范围为56.13~69.04mm、73.96~92.12mm、71.79~79.83°等，有助于临床上对骨缺损相关疾病的诊断和治疗。解剖参数在健康样本模型和SSM模型上的定性和定量对比分析的结果显示，线性参数值的平均偏差均小于6mm，角度参数值的平均偏差均小于2.5°，统计学检验的结果显示所有解剖参数的p-value>0.05，证明膝关节SSM预测模型与真实健康模型在几何外形上不具有统计学差异。 结论：该研究得出了健康膝关节解剖参数的参考范围，并证明了膝关节SSM模型在外形上与真实健康模型具有一致性，该结果可以作为基于SSM设计膝关节假体的依据。
Purpose: The purpose of this study was to obtain the range of anatomical parameters of healthy knee joints in Chinese men and to validate the statistical shape model(SSM) based on the healthy knee is validated in terms of geometric shape, which provides a basis for the design of knee SSM-based prosthesis. Methods: CT images (n>100) of healthy male knee joints were acquired to build 3D knee joint models, each model will be selected as the target sample, and the remaining models will be used as training set for Principal Component Analysis(PCA), and the obtained knee SSM will be fitted to the target sample model to predict the SSM prediction model. The same anatomical measurement points (6 on the femur, 8 on the patella, and 6 on the tibia) were marked on the sample model and SSM prediction model, respectively, and 20 anatomical parameters (17 linear and 3 angular parameters) were derived. The values of the anatomical parameters were statistically tested using independent samples t-test (parametric validation) and Mann-Whitney U-test (non-parametric validation), and the validity of the SSM in terms of geometric profile was demonstrated if the resulting p-values were all greater than 0.05. Results: The principal components vi obtained from PCA indicated different anatomical variability in the training set model, such as principal component 1 indicating volume differences in the training set model, as manifested by overall enlargement and reduction of the knee bone component, and principal components 2 and 3 indicating different proportional relationships in size and a certain change in knee flexion rotation, respectively. In this study, the ranges of anatomical parameters of the healthy right knee joint in Chinese males were derived, such as the reference ranges of LAP, TEA and ∠MA of 56.13~69.04mm, 73.96~92.12mm and 71.79~79.83°, which are helpful for the clinical diagnosis and treatment of bone defect-related diseases. The results of qualitative and quantitative comparative analysis of anatomical parameters on the healthy sample model and the SSM model showed that the mean deviation of linear parameter values was less than 6 mm and the mean deviation of angular parameter values was less than 2.5°. The results of statistical tests showed p-value>0.05 for all anatomical parameters, proving that the knee SSM prediction model is not statistically different from the true health model in terms of geometric shape. Conclusion: This study derived a reference range of anatomical parameters for the healthy knee and demonstrated that the knee SSM model is consistent with the real healthy model in terms of shape, and the results can be used as a basis for designing knee prostheses based on SSM.