Evaluation of functional methods of joint centre determination for quasi-planar movement

Meng, Lin and Childs, Craig and Buis, Arjan (2019) Evaluation of functional methods of joint centre determination for quasi-planar movement. PLOS One, 14 (1). e0210807. ISSN 1932-6203 (https://doi.org/10.1371/journal.pone.0210807)

[thumbnail of Meng-etal-PLOS-One-2019-Evaluation-of-functional-methods-of-joint-centre-determination]
Text. Filename: Meng_etal_PLOS_One_2019_Evaluation_of_functional_methods_of_joint_centre_determination.pdf
Final Published Version
License: Creative Commons Attribution 4.0 logo

Download (1MB)| Preview


Functional methods identify joint centres as the centre of rotation (CoR) of two adjacent movements during an ad-hoc movement. The methods have been used for functionally determining hip joint centre in gait analysis and have revealed advantages compared to predictive regression techniques. However, the current implementation of functional methods hinders its application in clinical use when subjects have difficulties performing multi-plane movements over the required range. In this study, we systematically investigated whether functional methods can be used to localise the CoR during a quasi-planar movement. The effects of the following factors were analysed: the algorithms, the range and speed of the movement, marker cluster location, marker cluster size and distance to the joint centre. A mechanical linkage was used in our study to isolate the factors of interest and give insight to variation in implementation of functional methods. Our results showed the algorithms and cluster locations significantly affected the estimate results. For all algorithms, a significantly positive relationship between CoR errors and the distance of proximal cluster coordinate location to the joint centre along the medial-lateral direction was observed while the distal marker clusters were best located as close as possible to the joint centre. By optimising the analytical and experimental factors, the transformation algorithms achieved a root mean square error (RMSE) of 5.3 mm while the sphere fitting methods yielded the best estimation with an RMSE of 2.6 mm. The transformation algorithms performed better in presence of random noise and simulated soft tissue artefacts.