Bakke Duncan, Zhang Ju, Hislop-Jambrich Jacqui, Besier Thor
Auckland Bioengineering Institute, University of Auckland, Auckland, New Zealand; FormusLabs, Auckland, New Zealand.
Clinical Applications Research Centre, Canon Medical, Sydney, Australia.
J Biomech. 2023 Jan;147:111418. doi: 10.1016/j.jbiomech.2022.111418. Epub 2022 Dec 24.
Accurate estimation of the hip joint centre (HJC) location is critical for modelling the kinematics and kinetics of the lower limb. Regression equations are commonly used to predict the HJC from anatomical landmarks on the pelvis, such as those published by Tylkowski et al., Andriacchi et al., Bell et al., and Seidel et al. Using a population of 159 CT-segmented pelvises, we assessed the accuracy of these methods as originally reported, and refined their parameters based on our larger cohort. We found the Tylkowski, Bell, and Seidel methods had mean Euclidean errors of 22.5, 26.4, and 17.9 mm, respectively. With new parameters for each method 'back-calculated' from our pelvic population, each method's error was reduced by an average of 69 %, with mean absolute errors of 7.9, 6.6, and 5.9 mm, respectively. For all methods, error has been reduced to below 1 cm, well below published levels for pelvic landmark estimation methods. These results highlight the need to validate and re-calibrate joint centre prediction methods on large, representative datasets to account for natural morphological variations.
准确估计髋关节中心（HJC）的位置对于模拟下肢的运动学和动力学至关重要。回归方程通常用于根据骨盆上的解剖标志预测HJC，例如Tylkowski等人、Andriacchi等人、Bell等人以及Seidel等人发表的方程。我们使用159个CT分割骨盆的群体，评估了这些方法最初报告时的准确性，并根据我们更大的队列对其参数进行了优化。我们发现Tylkowski、Bell和Seidel方法的平均欧几里得误差分别为22.5、26.4和17.9毫米。根据我们的骨盆群体“反算”出每种方法的新参数后，每种方法的误差平均降低了69%，平均绝对误差分别为7.9、6.6和5.9毫米。对于所有方法，误差已降至1厘米以下，远低于骨盆标志估计方法的已发表水平。这些结果凸显了在大型代表性数据集上验证和重新校准关节中心预测方法以考虑自然形态变异的必要性。
