Abstract
Introduction
Soft tissue artefact (STA) affects the kinematics retrieved with skin marker-based motion capture, and thus influences the outcomes of biomechanical models that rely on such kinematics. To date, compensation for STA remains an unsolved challenge due to its complexity. Factors include its dependency on subject, on motion activity and on skin-marker configuration, its non-linearity over the movement cycle, and the scarcity of reference in-vivo estimations. The objective of this study was extending the existing knowledge of the effects of STA on the kinematics of the hip joint and on the hip joint center location, by quantifying them for a sample total hip arthroplasty (THA) population, for a broader range of activities of daily living (ADLs).
Methods
Four activities of daily living (overground gait, stairs descent, chair rise and putting on socks) were measured simultaneously with optical motion capture (MC) at 100 Hz and with a movable single-plane video-fluoroscopy system (VF) at 25 Hz, for fifteen patients with successful total hip arthroplasty (THA). The joint segment positions were computed by least-square fitting for MC and by semi-automatic 2D/3D registration for VF. Anatomical coordinate systems were defined for each joint segment based on skin markers location at a reference standing position. Errors induced by STA on the retrieved joint motion were computed as the difference between MC-based kinematics and the reference VF-based kinematics. Statistical analysis was carried out to determine the whether the differences between the kinematics obtained with the two methods were significant.
Results
MC underestimated the ROM of the hip joint for all activities. The ROM for the flexion-extension was underestimated by on average 4.1°, 6.5°, 8.0° and 6.9° for gait, stair decent, chair rise and putting on socks respectively. Overall, during dynamic activities the hip joint was less flexed, more adducted and more internally rotated as retrieved using MC, compared to VF. The flexion angle was underestimated by MC during late stance phase and early swing phase for both gait and stairs descent. The internal rotation of the hip was overestimated by MC throughout the whole cycle of each activity. MC error for the thigh was larger than the MC error for the pelvis. MC errors above 1 cm were observed for the location of the hip joint center, with higher values for the cranial-caudal direction.
Discussion
Reduced ROM supports the notion that skin-sliding is a major contributor to STA. The underestimation of hip flexion appeared to be positively correlated to the degree of flexion. Larger skin deformation and sliding occurring for the thigh at higher hip flexion angles may produce the observed patterns of MC error for FE. The CC error was possibly due to inertial effects, and was more pronounced for the stair descent due to larger vertical movement and acceleration. This study led further bases for the activity-dependent correction of STA. This has the potential of improving the accuracy of motion capture and boosting its use for motion analysis as opposite to video-fluoroscopy techniques, which deliver radiation dose to the subjects.
