Comparison of muscle path predictions using OpenSim and a novel geodesic model

Abstract-Text (inkl. Referenzen und Bildunterschriften)
Introduction

Muscle forces in musculoskeletal modeling rely on the force-length relationship. Accurate muscle path prediction is crucial, typically assumed as straight lines and deviated using via points or wrapping surfaces. However, the modelled paths often exhibit inconsistencies and bone penetrations. To address this, we introduced a novel geodesic muscle path model[1,2].

Objective

Our goal is to predict realistic and consistent muscle paths over multiple bony surfaces using geodesics. We aim 1) to achieve muscle paths with zero bone penetration, 2) to prevent jumps of muscle paths and 3) to compare our predictions against those of the established OpenSim framework and against experimental muscle moment arms[3].

Method and Result

We approach the geodesic problem as a constrained optimisation with specified boundary conditions at the muscle origin and insertion points. Notably, our method guarantees zero penetration, contact and leave the bone surface tangentially and allow the incorporation of desired properties in the objective function, such as ensuring a smooth change in muscle length.

To test our framework, we developed several multi-body models with increasing complexity and number of muscles[1,4]. For each model, the geodesic method predicts consistent muscle paths and shows improvement compared to current solutions in OpenSim, with no bone penetration and prevent jumps. Both geodesic and OpenSim models were in close agreement with the experimental moment arm data.

Acknowledgment

This work was partly funded by the Deutsche Forschungsgemeinschaft (DFG) – SFB 1483 – Project-ID 442419336, EmpkinS and the Centre for Biomedical Technologies (QUT). PP acknowledges support from the Australian Research Council (IC190100020).

[1] Penner et al. Multibody System Dynamics 56 (2022)

[2] Scholz et al. Multibody System Dynamics 36 (2016)

[3] Hik et al. Journal of Anatomy 234 (2019)

[4] Audenaert et al. Computer Methods and Programs in Biomedicine 92 (2008)