Estimated landmark calibration of biomechanical models for inverse kinematics

Trinler, U and Baker, RJ 2017, 'Estimated landmark calibration of biomechanical models for inverse kinematics' , Medical Engineering & Physics .

Full text not available from this repository.

Abstract

Inverse kinematics is emerging as the optimal method in movement analysis to fit a multi-segment biomechanical model to experimental marker positions. A key part of this process is calibrating the model to the dimensions of the individual being analysed which requires scaling of the model, pose estimation and localisation of tracking markers within the relevant segment coordinate systems. The aim of this study is to propose a generic technique for this process and test a specific application to the OpenSim model Gait2392. Kinematic data from 10 healthy adult participants were captured in static position and normal walking. Results showed good average static and dynamic fitting errors between virtual and experimental markers of 0.8 cm and 0.9 cm, respectively. Highest fitting errors were found on the epicondyle (static), feet (static, dynamic) and on the thigh (dynamic). These result from inconsistencies between the model geometry and degrees of freedom and the anatomy and movement pattern of the individual participants. A particular limitation is in estimating anatomical landmarks from the bone meshes supplied with Gait2392 which do not conform with the bone morphology of the participants studied. Soft tissue artefact will also affect fitting the model to walking trials.

Item Type: Article
Uncontrolled Keywords: Biomechanical modelling, Human movement analysis, Inverse kinematics
Schools: Schools > School of Health Sciences
Journal or Publication Title: Medical Engineering & Physics
Publisher: Elsevier
ISSN: 1350-4533
Related URLs:
SWORD Depositor: Publications Router
Depositing User: Publications Router
Date Deposited: 21 Nov 2017 12:31
Last Modified: 21 Nov 2017 12:32
URI: http://usir.salford.ac.uk/id/eprint/44403

Actions (login required)

Edit record (repository staff only) Edit record (repository staff only)