(D-135) Investigating the Effects of STA on Anatomical Frame Estimates for Motion Capture

Introduction:

Human biomechanical modelling has many applications, ranging from space exploration to rehabilitation [1], [2]. Optical motion capture (OMC) is widely used to provide the data needed to create or customize models. However, OMC is sensitive to physical obstructions [3] and restricts data collection to the laboratory. On the other hand, inertial motion capture (IMC), which employs inertial measurement units (IMUs), is a cost-effective alternative [4] that can be used outside of a laboratory setting [5], [6], [7]. Also, while OMC requires a minimum of three markers per body segment, IMC requires only one IMU per body segment.

While IMC has its own unique drawbacks like ferromagnetic interference [8], both IMC and OMC are affected by soft tissue artifacts (STAs). In STA, the soft tissue that lies between the marker or IMU and bone shifts during motion, creating error [9]. STA’s effects on OMC have been extensively researched, and several studies describe significant STA on the thigh [10], [11]. However, the effect of STA on IMC data remains relatively understudied. IMC also lacks conventions and accuracy standards for defining anatomical frames. Given the numerous applications of extra-laboratory motion capture, further study is needed. Here, a simulation study is conducted using data from Cereatti, et al. [9] to understand how STAs impact marker position error, leading to error in instantaneous anatomical frame estimates.


Materials and Methods:
OpenSim [12] is an open-source biomechanical modeling platform that either generates virtual motion or analyzes actual motion from optical motion capture data. In recent years, the software has been used to study motions ranging from the gait pattern of an osseointegrated transfemoral amputee [13] to the backswing of a golf player [14]. Here, the gait2354_simbody model (hereafter referred to as Gait2354) is utilized because it includes a comprehensive set of lower body muscles but is simplified enough to reduce computation costs. Others have used this model to describe motor adaptation to exoskeleton use [15] or predict astronauts’ musculoskeletal performance in spacesuits [1]. In this study, a customized Gait2354 model is used to produce unique trials and generate inverse kinematics results for every time-step in each trial. These results are then imported into MATLAB [16] to conduct data analysis and visualization.


To investigate the effect of STA on anatomical reference frames, datasets included in [9] and [10] are used to add varying magnitudes of STA error to the marker data included with the Gait 2354 model. Model scaling, inverse kinematics, static optimization, and body kinematics analyses are performed using OpenSim and including all markers at equal weight. The resulting position data is exported to MATLAB, where the relationships between position error over time, position error versus Euler angle error, and position difference versus rotation difference (1) are explored and analyzed.



Results, Conclusions, and Discussions:: In comparing femur position against time for models with varying magnitudes of STA, models with increased STA generally overestimate the position of the femur (Figure 1). This trend is positively correlated with increased STA. In the model with reduced STA, the position of the femur is underestimated compared to the normal model. When comparing rotation and position errors, a clear trend of increased position error leading to increased rotation error is observed (Figure 2, Table 1).

Intuitively, greater STAs generate larger errors in anatomical frame estimates. Further research on STAs is needed to fully describe its effect on both optical and inertial motion capture with respect to anatomical frame estimation accuracy. A more complete understanding of STAs will help define standards for anatomical reference frames for inertial motion capture. Unlike previous work, this investigation is acutely interested in how STAs affect instantaneous anatomical frame estimates for an individual body segment (i.e., thigh). This focus is unique as most past work has focused on how STAs affect marker positions or subsequent joint angles, which are a nonlinear combination of the errors from two body segments. Future work will utilize these rotation difference findings to develop standardization criteria for establishing anatomical frames for inertial motion capture, which is conspicuously lacking conventions [6]. Since STAs are the most significant source of error in optical motion capture, this study leverages that fact to derive a minimum level of anatomical frame estimation accuracy for inertial motion capture. Future work includes incorporating data from participants with more varied demographics and morphologies engaging in different ambulation speeds.


Acknowledgements: This work was supported by National Science Foundation Grant 2049044

References:
A. Diaz Artiles, et al., “Musculoskeletal injuries and minor trauma in space: incidence and injury mechanisms in U.S. astronauts,” Aviat. Space Environ. Med., vol. 37, no. 8, Jul. 2012, doi: 10.2514/6.2012-3548.
M. Pizzolato, L. Reggiani, L. Modenese, and D. Lloyd, “Real-time inverse kinematics and inverse dynamics for lower limb applications using OpenSim.” Comput. Methods Biomech. Biomed. Eng., vol. 20, no. 4, pp. 436–445, Sept. 2016, doi: 10.1080/10255842.2016.1240789.
J. Lee, S. Shin, G. Ghorpade, T. Akbas and J. Sulzer, "Sensitivity comparison of inertial to optical motion capture during gait: implications for tracking recovery," 2019 IEEE 16th International Conference on Rehabilitation Robotics (ICORR), Toronto, ON, Canada, pp. 139-144, 2019, doi: 10.1109/ICORR.2019.8779411.
M. Al Borno, et al., “OpenSense: An open-source toolbox for inertial-measurement-unit-based measurement of lower extremity kinematics over long durations,” BioMed Central Ltd, vol. 19, no. 22, Feb. 2022, doi: 10.1186/s12984-022-01001-x.
V. Davila, et al., “Intraprocedural ergonomics of vascular surgeons,” J. of Vasc. Surg., vol. 73, no. 1, pp. 301–308, Jan. 2021, doi: 10.1016/j.jvs.2020.04.523.
R. Vitali, et al., “Body-worn IMU array reveals effects of load on performance in an outdoor obstacle course,” PLoS ONE, vol. 14, no. 3, Mar. 2019, doi: 10.1371/journal.pone.0214008.
W. Wang and P. Adamczyk, “Analyzing gait in the real world using wearable movement sensors and frequently repeated movement paths,” IEEE Sens. J., vol. 19, no. 8, Apr. 2019, doi: 10.3390/s19081925.
G. Ligorio and A. Sabantini, “Dealing with Magnetic Disturbances in Human Motion Capture: A Survey of Techniques,” Micromachines, vol. 7, no. 3, Mar. 2016, doi: 10.3390/mi7030043.
A. Cereatti, et al., “Standardization proposal of soft tissue artefact description for data sharing in human motion measurements,” J. of Biomech., vol. 62, pp. 5–13, Sept. 2017, doi: 10.1016/j.jbiomech.2017.02.004.
A. Barré, J. Thiran, M. Jolles., N. Theumann, and K. Aminian, “Soft tissue artifact assessment during treadmill walking in subjects with total knee arthroplasty,” IEEE Trans. Biomed. Eng. vol. 60, no. 11, pp. 3131–3140, Nov. 2013, doi: 10.1109/TBME.2013.2268938.
A. Peters, et al., “Quantification of soft tissue artifact in lower limb human motion analysis: A systematic review,” Gait Posture, vol. 31, no. 1, pp. 1–8, Jan. 2020, doi: 10.1016/j.gaitpost.2009.09.004.
S. Delp, et al., “OpenSim: Open-source software to create and analyze dynamic simulations of movement,” IEEE Trans. on Biomed. Eng., vol. 54, no. 11, pp. 1940-1950, Nov. 2007, doi: 10.1109/TBME.2007.901024.
V. Raveendranathan and R. Carloni, "Musculoskeletal Model of an Osseointegrated Transfemoral Amputee in OpenSim," 2020 8th IEEE RAS/EMBS International Conference for Biomedical Robotics and Biomechatronics (BioRob), New York, NY, USA, 2020, pp. 1196-1201, doi: 10.1109/BioRob49111.2020.9224422.
S. Mahadas, K. Mahadas, and G. Hung, “Biomechanics of the golf swing using OpenSim.” Comput. Biol. Med., vol. 105, pp. 39-45, Feb. 2019, doi: 10.1016/j.compbiomed.2018.12.002.
F. Haufe, et al., “Learning to walk with a wearable robot in 880 simple steps: a pilot study on motor adaptation,” J. NeuroEng. Rehabilitation, vol. 18, no. 1, Nov. 2021, doi: /10.1186/s12984-021-00946-9.
The MathWorks, Inc. (2023). MATLAB version: 9.9.0 (R2022b). Accessed: July 05, 2023. Available: https://www.mathworks.com.