Enhancing Dynamic Consistency of Multimodal Motion Data in Musculoskeletal Simulation

Wechsler I, Wolf A, Wartzack S, Miehling J (2022)


Publication Language: English

Publication Type: Conference contribution, Abstract of a poster

Publication year: 2022

Event location: Porto PT

Abstract

Introduction
Digital human models can be used to determine biomechanical parameters (e.g. muscle or joint reaction forces) virtually and in a non-invasive way [1]. Hereby, subject-specific models lead to more reliable simulation results in this context [2]. Moreover, even though current sensor and measuring systems can achieve high-quality data, the measurements are still error-prone and subject to uncertainties such as noise or jumps. Furthermore, kinematic and dynamic inconsistencies between the measured data and the motion capabilities of the human body can occur. We assume that motion tracking accuracy and consistency will be increased when multimodal data, e.g., some combination of position, orientation, EMG, or surface data, is considered. Therefore, our goal is to research and develop methods for filtering and analysing multimodal motion measurement data in conjunction with individualized digital human models. In this paper we want to discuss available approaches.

Methods
Subject-specific individualization of the applied musculoskeletal human models in multiple domains (anthropometry, mobility, strength) is to be achieved through a combination of subject-specific manual measurements and population data [3]. Further, we aim to develop a method for kinematically consistent and precise motion tracking. The standard method, Inverse Kinematics, minimises the difference between measured and virtual marker and coordinate positions in a least squares approach to compute generalized coordinate trajectories [4]. The novel tracking algorithm should include different types of data, which are weighted in relation to each other. This way the observed movement information from different data sources should be transferred to the digital model in order to get the most accurate and consistent simulation results possible. Nevertheless, the problem regarding the dynamic consistencies between the motion of the model and the measurements still remains. For this we also want to investigate methods to enhance dynamic consistency and further develop a dynamic tracking method that is able to generate kinematically and also dynamically consistent motion data. Therefore, dynamic components (e.g. muscle activation, accelerations) are to be included in the tracking method.

Results
There are various approaches available in literature for enhancing dynamic consistency (cf. figure 1). The EMG-informed forward simulation approach, for example, tracks measured marker trajectories while replicating EMG measurements to obtain dynamic consistency [5-7]. Controller-based solution approaches optimise measured trajectories through forward simulation with a time-limited observation interval to enhance dynamic consistency [8]. Also, trajectory optimization using direct collocation and a Kalman filter have already been used to achieve dynamically consistent movement simulations [9,10].

Discussion
In the next step, a systematic literature review will be conducted based on the mentioned results, in order to identify and analyse all relevant dynamic tracking solutions regarding multimodal motion data. Main focus will be the classification into different solution approaches. Moreover, input and output variables, the used model and the regarded dynamic component will be explored. Based on these results the solution approach for our application of multimodal motion data is to be set up. This new musculoskeletal simulation approach is then to be evaluated regarding accuracy and usability and to be benchmarked against the standard methods of biomechanical motion capture and analysis.

References
1. Seth et al, Computational Biol, 14(7), 2018.
2. Lund et al, Internat Biomech, 2:1-11, 2015.
3. Miehling, Comput Methods Biomech Biomed Engin, 22(15):1209-1218, 2019.
4. Delp et al, IEEE Trans Biomed Eng, 54: 1940-1950, 2007.
5. Sartori et al, PLoS One, 7(12), 2012.
6. Bailly et al, Front Bioeng Biotechnol, 2021.
7. Moissenet et, Front. Neurorobot, 2019.
8. Thelen et al, J Biomech, 39:1107-1115, 2006.
9. Dembia et al, PLoS Comput Biol, 16(12), 2020.
10. Bonnet et al, J Biomech, 62:140-147, 2017.

Acknowledgements
This work was (partly) supported by the Deutsche Forschungs-gemeinschaft (DFG, German Research Foundation) under Grant SFB 1483–Project-ID 442419336.


Authors with CRIS profile

Related research project(s)

How to cite

APA:

Wechsler, I., Wolf, A., Wartzack, S., & Miehling, J. (2022, June). Enhancing Dynamic Consistency of Multimodal Motion Data in Musculoskeletal Simulation. Poster presentation at 27th Congress of the European Society of Biomechanics, Porto, PT.

MLA:

Wechsler, Iris, et al. "Enhancing Dynamic Consistency of Multimodal Motion Data in Musculoskeletal Simulation." Presented at 27th Congress of the European Society of Biomechanics, Porto 2022.

BibTeX: Download