Analytical study of the effects of soft tissue artefacts on functional techniques to define axes of rotation

Abstract

The accurate location of the main axes of rotation (AoR) is a crucial step in many applications of human movement analysis. There are different formal methods to determine the direction and position of the AoR, whose performance varies across studies, depending on the pose and the source of errors. Most methods are based on minimizing squared differences between observed and modelled marker positions or rigid motion parameters, implicitly assuming independent and uncorrelated errors, but the largest error usually results from soft tissue artefacts (STA), which do not have such statistical properties and are not effectively cancelled out by such methods. However, with adequate methods it is possible to assume that STA only account for a small fraction of the observed motion and to obtain explicit formulas through differential analysis that relate STA components to the resulting errors in AoR parameters. In this paper such formulas are derived for three different functional calibration techniques (Geometric Fitting, mean Finite Helical Axis, and SARA), to explain why each technique behaves differently from the others, and to propose strategies to compensate for those errors. These techniques were tested with published data from a sit-to-stand activity, where the true axis was defined using bi-planar fluoroscopy. All the methods were able to estimate the direction of the AoR with an error of less than 5°, whereas there were errors in the location of the axis of 30-40mm. Such location errors could be reduced to less than 17mm by the methods based on equations that use rigid motion parameters (mean Finite Helical Axis, SARA) when the translation component was calculated using the three markers nearest to the axis.

Technical note: sensitivity analysis of the SCoRE and SARA methods for determining rotational axes during tibiofemoral movements using optical motion capture

“…A correction equation led to a sum of the root mean square error of between 0.6 and 0.8 degrees for the SARA method [16]. In addition, De Rosario et al [7] presented a mathematical model of soft tissue artefacts propagation to the position and direction of variable and fixed axes as calculated by three methods. One of these methods was the SARA method.…”

“…The reliability of the SCoRE and SARA methods including their application by Boeth et al [5] have been studied several times [7, 16, 25, 26]. A high reliability (Intraclass correlation coefficient > 0.8) and no significant differences between five different observers who placed the marker set and between measure days was found in functional femur and tibia length (distances between the centers of the axis of rotation and the hip center or ankle center) calculated based on the SCoRE and SARA methods [25].…”

“…There are various methods to estimate the joint center of rotation and axis of rotation (i.e. [6,7,11,14]). In this technical note the symmetrical center of rotation estimation (SCoRE) [8] and the symmetrical axes of rotation approach (SARA) [9] which are implemented in the software of the motion capture system Vicon (VICON able to evaluate the functional highly significant dynamic movement, such a method is very useful.…”

A methodology for the customization of hinged ankle-foot orthoses based on in vivo helical axis calculation with 3D printed rigid shells

“…In the frontal plane, the orientation of the MHA resulted approximately parallel to the intermalleolar line segment. 30 Even though results compare well, the accuracy of joint axis of rotation based on IHA theory still presents some critical aspects, to be examined in depth as pointed out in the recent literature, 31,32 especially regarding the precision in the evaluation of axode location for wide movements. Alternative techniques such as finite helical axis, symmetrical axis of rotation approach (SARA) 32 or geometric fitting (GF) 33 can provide similar or even better results for motion analysis based on stereophotogrammetry.…”

Hip and knee kinematics of the forward lunge one year after unicondylar and total knee arthroplasty

“…Motion capture data processing was performed in Visual 3D V5 software (Cmotion Inc., Germantown, US), filtering was applied using a Butterworth filter with a cut-off frequency set at 6Hz. Knee and hip joint axis were computed using a functional joint calibration procedure [De Rosario et al 2017].…”