Skip to main content
SpringerLink
Log in

Subject-specific trunk segmental masses prediction for musculoskeletal models using artificial neural networks

  • Original Article
  • Published:
Medical & Biological Engineering & Computing Aims and scope Submit manuscript

Abstract

Accurate determination of body segment parameters is crucial for studying human movement and joint forces using musculoskeletal (MSK) models. However, existing methods for predicting segment mass have limited generalizability and sensitivity to body shapes. With recent advancements in machine learning, this study proposed a novel artificial neural network-based method for computing subject-specific trunk segment mass and center of mass (CoM) using only anthropometric measurements. We first developed, trained, and validated two artificial neural networks that used anthropometric measurements as input to predict body shape (ANN1) and tissue mass (ANN2). Then, we calculated trunk segmental mass for two volunteers using the predicted body shape and tissue mass. The body shape model (ANN1) was tested on 279 subjects, and maximum deviation between the predicted body shape and the original was 28 mm. The tissue mass model (ANN2) was evaluated on 223 subjects, which when compared to ground truth data, had a mean error of less than 0.51% in the head, trunk, legs, and arms. We also compared the two volunteer’s trunk segment mass with experimental data and found similar trend and magnitude. Our findings suggested that the proposed method could serve as an effective and convenient tool for predicting trunk mass.

Graphical Abstract

This is a preview of subscription content, log in via an institution to check access.

Access this article

Log in via an institution

Price excludes VAT (USA)
Tax calculation will be finalised during checkout.

Instant access to the full article PDF.

Institutional subscriptions

Fig. 1
Fig. 2
Fig. 3
Fig. 4
Fig. 5

Similar content being viewed by others

References

  1. Rao G, Amarantini D, Berton E, Favier D (2006) Influence of body segments’ parameters estimation models on inverse dynamics solutions during gait. J Biomech 39:1531–1536. https://doi.org/10.1016/j.jbiomech.2005.04.014

    Article  PubMed  Google Scholar 

  2. Narang YS, Murthy Arelekatti VN, Winter AG (2016) The effects of the inertial properties of above-knee prostheses on optimal stiffness, damping, and engagement parameters of passive prosthetic knees. J Biomech Eng 138. https://doi.org/10.1115/1.4034168

  3. Pearsall DJ, Costigan PA (1999) The effect of segment parameter error on gait analysis results. Gait Posture 9:173–183. https://doi.org/10.1016/s0966-6362(99)00011-9

    Article  CAS  PubMed  Google Scholar 

  4. Silva MPT, Ambrósio JAC (2004) Sensitivity of the results produced by the inverse dynamic analysis of a human stride to perturbed input data. Gait Posture 19:35–49. https://doi.org/10.1016/S0966-6362(03)00013-4

    Article  PubMed  Google Scholar 

  5. Fritz J, Kröll J, Schwameder H (2019) Influence of body segment parameter estimation on calculated ground reaction forces in highly dynamic movements. J Biomech 84:11–17. https://doi.org/10.1016/j.jbiomech.2018.12.008

    Article  PubMed  Google Scholar 

  6. Köhler H-P, Schüler A, Quaas F et al (2024) The influence of body segment estimation methods on body segment inertia parameters and joint moments in javelin throwing. Comput Methods Biomech Biomed Engin 27:267–275. https://doi.org/10.1080/10255842.2023.2181039

    Article  PubMed  Google Scholar 

  7. Hanavan EP (1964) A Mathematical Model of the Human Body, Aerospace Medical Research Laboratory, Wright-Petterson Air Force Base, Ohio, USA

  8. Sheets AL, Corazza S, Andriacchi TP (2010) An automated image-based method of 3D subject-specific body segment parameter estimation for kinetic analyses of rapid movements. J Biomech Eng 132:011004. https://doi.org/10.1115/1.4000155

    Article  PubMed  Google Scholar 

  9. Martin PE, Mungiole M, Marzke MW, Longhill JM (1989) The use of magnetic resonance imaging for measuring segment inertial properties. J Biomech 22:367–376. https://doi.org/10.1016/0021-9290(89)90051-1

    Article  CAS  PubMed  Google Scholar 

  10. Huang HK, Wu SC (1976) The evaluation of mass densities of human body in vivo from CT scans. Comput Biol Med 6:337–343. https://doi.org/10.1016/0010-4825(76)90070-6

    Article  CAS  PubMed  Google Scholar 

  11. Durkin JL, Dowling JJ (2006) Body segment parameter estimation of the human lower leg using an elliptical model with validation from DEXA. Ann Biomed Eng 34:1483–1493. https://doi.org/10.1007/s10439-006-9088-6

    Article  PubMed  Google Scholar 

  12. Norton J, Donaldson N, Dekker L (2002) 3D whole body scanning to determine mass properties of legs. J Biomech 35:81–86. https://doi.org/10.1016/S0021-9290(01)00161-0

    Article  PubMed  Google Scholar 

  13. Smith SHL, Bull AMJ (2018) Rapid calculation of bespoke body segment parameters using 3D infra-red scanning. Med Eng Phys 62:36–45. https://doi.org/10.1016/j.medengphy.2018.10.001

    Article  PubMed  Google Scholar 

  14. Kudzia P, Jackson E, Dumas G (2022) Estimating body segment parameters from three-dimensional human body scans. PLoS ONE 17:e0262296. https://doi.org/10.1371/journal.pone.0262296

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  15. Liu T, Khalaf K, Hebela N et al (2021) Prediction of human male trunk mass distribution using anthropometric measurements: A feasibility study. J Biomech 122:110437. https://doi.org/10.1016/j.jbiomech.2021.110437

    Article  PubMed  Google Scholar 

  16. Spitzer V, Ackerman MJ, Scherzinger AL, Whitlock D (1996) The visible human male: a technical report. J Am Med Inform Assoc 3:118–130. https://doi.org/10.1136/jamia.1996.96236280

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  17. Pearsall DJ, Reid JG, Livingston LA (1996) Segmental inertial parameters of the human trunk as determined from computed tomography. Ann Biomed Eng 24:198–210

    Article  CAS  PubMed  Google Scholar 

  18. Varol G, Ceylan D, Russell B et al (2018) BodyNet: Volumetric Inference of 3D Human Body Shapes. In: Ferrari V, Hebert M, Sminchisescu C, Weiss Y (eds) Computer Vision – ECCV 2018. Lecture Notes in Computer Science, vol 11211. Springer, Cham, pp 20–38. https://doi.org/10.1007/978-3-030-01234-2_2

  19. Kocabas M, Athanasiou N, Black MJ (2020) VIBE: Video Inference for Human Body Pose and Shape Estimation. 2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR). IEEE, Seattle, WA, USA, pp 5252–5262

    Chapter  Google Scholar 

  20. Pishchulin L, Wuhrer S, Helten T et al (2017) Building statistical shape spaces for 3D human modeling. Pattern Recogn 67:276–286. https://doi.org/10.1016/j.patcog.2017.02.018

    Article  Google Scholar 

  21. Robinette KM, Daanen H, Paquet E (1999) The CAESAR project: a 3-D surface anthropometry survey. In: Second International Conference on 3-D Digital Imaging and Modeling (Cat. No.PR00062), pp 380–386. https://doi.org/10.1109/IM.1999.805368

  22. Clauser CE, McConville JT, Young JW (1971) Weight, volume, and center of mass segments of the human body. J Occup Environ Med 13:270

    Google Scholar 

  23. Dempster WT (1955) Space requirements of the seated operator: geometrical, kinematic, and mechanical aspects of the body with special reference to the limbs. Wright Air Development Center, Wright-Patterson Air Force Base, Ohio

    Google Scholar 

  24. Tian IY, Ng BK, Wong MC et al (2020) Predicting 3D body shape and body composition from conventional 2D photography. Med Phys 47:6232–6245. https://doi.org/10.1002/mp.14492

    Article  PubMed  Google Scholar 

  25. Zhu H, Zuo X, Wang S et al (2019) Detailed human shape estimation from a single image by hierarchical mesh deformation. 2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR). IEEE, Long Beach, CA, USA, pp 4486–4495

    Chapter  Google Scholar 

  26. Zhu S, Mok PY (2015) Predicting realistic and precise human body models under clothing based on orthogonal-view photos. Procedia Manuf 3:3812–3819. https://doi.org/10.1016/j.promfg.2015.07.884

    Article  Google Scholar 

  27. Anguelov D, Srinivasan P, Koller D et al (2005) SCAPE: shape completion and animation of people. ACM Trans Graph 24:408–416. https://doi.org/10.1145/1073204.1073207

    Article  Google Scholar 

  28. Alldieck T, Kassubeck M, Magnor M (2017) Optical flow-based 3d human motion estimation from monocular video. arXiv. http://arxiv.org/abs/1703.00177

  29. Huang Y, Bogo F, Lassner C et al (2017) Towards accurate marker-less human shape and pose estimation over time. In: 2017 International Conference on 3D Vision (3DV), pp 421–430. https://arxiv.org/abs/1707.07548

  30. Wuhrer S, Shu C (2013) Estimating 3D Human Shapes from Measurements. Mach Vis Appl 24:1133–1147. https://doi.org/10.1007/s00138-012-0472-y

    Article  Google Scholar 

  31. Mungiole M, Martin PE (1990) Estimating segment inertial properties: comparison of magnetic resonance imaging with existing methods. J Biomech 23:1039–1046. https://doi.org/10.1016/0021-9290(90)90319-x

    Article  CAS  PubMed  Google Scholar 

  32. Cheng CK, Chen HH, Chen CS et al (2000) Segment inertial properties of Chinese adults determined from magnetic resonance imaging. Clin Biomech (Bristol, Avon) 15:559–566. https://doi.org/10.1016/s0268-0033(00)00016-4

    Article  CAS  PubMed  Google Scholar 

  33. Makovey J, Naganathan V, Sambrook P (2005) Gender differences in relationships between body composition components, their distribution and bone mineral density: a cross-sectional opposite sex twin study. Osteoporos Int 16:1495–1505. https://doi.org/10.1007/s00198-005-1841-4

    Article  PubMed  Google Scholar 

  34. Speerin R, Needs C, Chua J et al (2020) Implementing models of care for musculoskeletal conditions in health systems to support value-based care. Best Pract Res Clin Rheumatol 34:101548. https://doi.org/10.1016/j.berh.2020.101548

    Article  PubMed  PubMed Central  Google Scholar 

  35. Shen Y-W, Yang Y, Liu H et al (2022) Biomechanical evaluation of intervertebral fusion process after anterior cervical discectomy and fusion: a finite element study. Front Bioeng Biotechnol 10. https://doi.org/10.3389/fbioe.2022.842382

  36. Imamura Y, Tanaka T, Suzuki Y et al (2011) Motion-Based-Design of Elastic Material for Passive Assistive Device Using Musculoskeletal Model. J Robot Mechatron 23:978–990. https://doi.org/10.20965/jrm.2011.p0978

Download references

Acknowledgements

This manuscript is based upon work supported by the Khalifa University of Science Technology [Award No FSU-2018-13].

Author information

Authors and Affiliations

  1. Human Performance Lab, Faculty of Kinesiology, University of Calgary, Calgary, AB, Canada

    Tao Liu

  2. Department of Mechanical Engineering, Khalifa University, Abu Dhabi, United Arab Emirates

    Marwan El-Rich

Authors
  1. Tao Liu
    View author publications

    You can also search for this author in PubMed Google Scholar

  2. Marwan El-Rich
    View author publications

    You can also search for this author in PubMed Google Scholar

Corresponding author

Correspondence to Tao Liu.

Ethics declarations

Conflict of interest

All authors declare that there is no conflict of interest.

Additional information

Publisher's Note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Appendix

Appendix

1.1 Evaluation of body shape neural network

The accuracy of the body shape model was evaluated by computing the error in principal component coefficient (PCC) between the ground truth and the predictions of ANN1. The results are shown in Table 3. The mean error of the 15 principal component coefficients ranged from -0.09 for PCC5 to 0.071 at PCC3, and the standard deviation varied from 0.168 at PCC1 to 1.046 at PCC3. Among the 15 coefficients, PCC1 exhibited the smallest minimum and maximum value of -0.643 and 0.525, respectively, followed by PCC4 with a minimum and maximum value of -1.007 and 0.811, respectively. The remaining PCC varied from around 2 to 4. The 25th and 75th percentiles showed similar values of approximately -0.5 and 0.5, respectively, for all PCC except for PCC1, which had values of -0.098 and 0.11.

Table 3 Statistics of principal component coefficient error between ground truth and the predicted test data (The error is defined as predicted data minus ground truth)
Full size table

To visually assess the deviation between the ground truth, reconstructed body shape and predicted body shape, 3D deviation contours were computed and visualized. The distances between each point of the ground truth body shape and the corresponding point on the predicted body shape were calculated and sorted in ascending order. The 3D deviation contours were then plotted at the distance threshold of 60%, 70%, 80%, 90%, and 100% using the test dataset (279 subjects) for comparison. In addition to the deviation between the ground truth and the predicted body shape (fourth column in Fig. 3), the deviation between the reconstructed body shape (represented by the first 15 principal shape variations) and the predicted body shape was also quantified (fifth column in Fig. 3).

When using the predicted body shape as a reference, both the ground truth and the compressed body shape showed a similar deviation contour, with the magnitude of the reconstructed body shape being smaller than that of the ground truth. As the threshold increased from 60% to 100%, both the reconstructed and ground truth body shapes showed an increasing maximum deviation and area with large deviation. More specifically, the ground truth showed a relatively small deviation area of around 17 mm at the 60% threshold, while the deviation increased to around 50 mm at 100% threshold. A similar trend was found for the reconstructed body shape, with a deviation area of around 7 mm at the 60% threshold and around 28 mm at the 100% threshold.

1.2 Evaluation of tissue mass neural network

Table 4 Statistics of segment percentage error between ground truth and the predicted test data (The error is defined as the difference between ground truth and predicted data)
Full size table

The error of segment percentage (head, arms, legs, trunk bone, trunk fat and trunk lean) between the ground truth and the predicted values was quantified on the test dataset (223 individuals). The statistics of the segment percentage error showed that mean error was relatively small and less than 0.51% (Table 4). The standard deviation of the segment percentage error varied from 0.13 in the trunk bone to 2.27 in trunk fat. The peak segment error values were smallest in the trunk bone with maximum and minimum values being -0.32% and 0.44%. This is followed by arms and head with minimum and maximum values of -1.84% and 3.22%, and -2.13% and 3.14%, respectively. The remaining peak segment error ranged from 5.87% to 9.35%. However, the 25th and 75th percentiles of segment error had relatively smaller values in the trunk bone (-0.09% and 0.08%), head (-0.47% and 0.35%), arms (-0.41% and 0.68%) but slightly larger values in the legs (-1.52% and 1.03%), trunk fat (-1.61% and 0.98%) and trunk lean (-0.87% and 1.83%).

Rights and permissions

Springer Nature or its licensor (e.g. a society or other partner) holds exclusive rights to this article under a publishing agreement with the author(s) or other rightsholder(s); author self-archiving of the accepted manuscript version of this article is solely governed by the terms of such publishing agreement and applicable law.

Reprints and permissions

About this article

Check for updates. Verify currency and authenticity via CrossMark

Cite this article

Liu, T., El-Rich, M. Subject-specific trunk segmental masses prediction for musculoskeletal models using artificial neural networks. Med Biol Eng Comput (2024). https://doi.org/10.1007/s11517-024-03100-4

Download citation

  • Received:

  • Accepted:

  • Published:

  • DOI: https://doi.org/10.1007/s11517-024-03100-4

Keywords

Search

Navigation