Accuracy and Reliability of Marker-Based Approaches to Scale the Pelvis, Thigh, and Shank Segments in Musculoskeletal Models

Abstract

Gait analysis together with musculoskeletal modeling is widely used for research. In the absence of medical images, surface marker locations are used to scale a generic model to the individual's anthropometry. Studies evaluating the accuracy and reliability of different scaling approaches in a pediatric and/or clinical population have not yet been conducted and, therefore, formed the aim of this study. Magnetic resonance images (MRI) and motion capture data were collected from 12 participants with cerebral palsy and 6 typically developed participants. Accuracy was assessed by comparing the scaled model's segment measures to the corresponding MRI measures, whereas reliability was assessed by comparing the model's segments scaled with the experimental marker locations from the first and second motion capture session. The inclusion of joint centers into the scaling process significantly increased the accuracy of thigh and shank segment length estimates compared to scaling with markers alone. Pelvis scaling approaches which included the pelvis depth measure led to the highest errors compared to the MRI measures. Reliability was similar between scaling approaches with mean ICC of 0.97. The pelvis should be scaled using pelvic width and height and the thigh and shank segment should be scaled using the proximal and distal joint centers.

Statistical shape modelling versus linear scaling: Effects on predictions of hip joint centre location and muscle moment arms in people with hip osteoarthritis

“…Virtual markers were placed on the OpenSim model to correspond with the experimental surface landmarks from the motion capture trials. Prior to scaling the pelvis in OpenSim, additional virtual markers were computed from the experimental motion capture data to enable the pelvis to be linearly scaled in three orthogonal directions ( Figure 2) (Kainz et al, 2017c). Scale factors were computed based on the ratio between experimental and virtual markers on the model.…”

“…Typically, musculoskeletal models are personalised by simple linear scaling of segment anthropometries from a generic musculoskeletal model (Kainz et al, 2017c). Individual segment dimensions can be determined using surface anatomical landmarks alone (O'Connor et al, 2018) or together with derived hip joint centre (HJC) locations (Kainz et al, 2017c). Inclusion of HJC locations in the scaling process has been shown to result in more precise estimates of pelvis dimensions compared to using surface anatomical markers alone (Kainz et al, 2017c).…”

“…Individual segment dimensions can be determined using surface anatomical landmarks alone (O'Connor et al, 2018) or together with derived hip joint centre (HJC) locations (Kainz et al, 2017c). Inclusion of HJC locations in the scaling process has been shown to result in more precise estimates of pelvis dimensions compared to using surface anatomical markers alone (Kainz et al, 2017c). Hip joint centre (HJC) locations should be estimated using either functional joint centre estimation methods in people with sufficient hip range of motion (ROM) (Piazza et al, 2001;Wu et al, 2002) or Harrington's regression equation for subjects with limited hip ROM (Kainz et al, 2015).…”

Reliability of four models for clinical gait analysis

“…Joint kinetics were calculated via Inverse Dynamics using segment mass and inertia parameters from [23]. Both IK models were scaled to each person using scale factors derived from surface marker positions and joint centres [25] (supplementary Table S1). For the marker placer task in OpenSim, only the anatomical landmar according to the experimental marker locations ( Fig.…”

Lower functioning patients demonstrate atypical hip joint loading before and following total hip arthroplasty for osteoarthritis

“…Musculoskeletal models are valuable for estimating internal joint loads that are difficult to measure in vivo, though the level of model complexity required to achieve reliable estimates of joint loading have remained contentious 7,13 . Typically, the bone dimensions of a generic musculoskeletal model are linearly scaled to conform to those of the patient using reflective markers placed on a participants' skin during motion capture experiments 14 . However, this approach has demonstrated large errors (up to 40 mm) in estimating the hip joint center (HJC) location in patients with hip pathology 15 where errors of this magnitude can result in a miscalculation of the resultant joint reaction force in the order of 0.5 times body weight 7 .…”