Ifsttar PhD subject

French version
Detailed form :

Title : External and internal statistical shape modelling for applications in biomechanics

Main host Laboratory - Referent Advisor TS2 - LBMC  -  BEILLAS Philippe      tél. : +33 472142371

Director of the main host Laboratory MITTON David  -

PhD Speciality Mécanique/Biomécanique

Axis of the performance contract 1 - COP2017 - Efficient transport and safe travel

Main location Bron

Doctoral affiliation UNIVERSITE CLAUDE-BERNARD-LYON 1

PhD school MEGA (MECANIQUE, ENERGETIQUE, GENIE CIVIL, ACOUSTIQUE)

Planned PhD supervisor WANG Xuguang  -  Université Gustave Eiffel  -  TS2 - LBMC

Planned PhD co-supervisor LAFON-JALBY Yoann  -    -

Planned financing Contrat doctoral  - Université Gustave Eiffel

Abstract

Context:

In the transportation field, vehicle automation requires solutions that provide information on occupant status, posture, activities, and more generally, any biomechanical characteristic that would be relevant to assess seating comfort and injury risk in case of accident. This potentially includes both internal and external information. It should be possible to generate this information with very partial sensor data (monitoring concept). Before that, the same type of data should be used to design new interior concepts and layouts, for both comfort and safety. For this, customizable/parametric models of the human body using deformable finite element and rigid multi-body methods are needed to assess seating comfort (Wang et al., 2019) as well as safety (Grébonval et al., 2021).

In health, a large amount of imaging data is available today and has enabled the development of segmentation algorithms based on learning and statistical shape models of bones. However, these data are typically obtained in the supine position and do not include the whole body. For example, prediction of spinal curvature from the outer shape is an active field of research (Nérot et al., 2016).

In sports and clothing, this type of model combining a statistical articulated envelope with a model of anatomical internal joints is helpful in design for sports and leisure retail companies (e.g. ongoing project with a sports equipment manufacturer).

In any case, if external and internal models exist, there is no statistical shape model of the of the whole body that includes external and internal structures and that can be realistically positioned. Such a model would have many applications in predicting most likely postures or internal characteristics from partial data (external patient measurements, real-time data in a vehicle for comfort or safety, markerless measurements, etc.).

However, the development of such a model presents a number of challenges and gaps.

For the skeleton, existing results are partial. Research efforts are focused on the creation of local statistical shape models (femur, knee joint, etc.). The LBMC contributes to this effort with the creation of statistical shape models (pelvis and femur, Savonnet et al., 2019, but also thorax, lower limb, skull). But the assembly of these parts from different populations/data sources remains a scientific challenge to ensure consistency and relative positioning of anatomical structures. A strong potential exists however with open databases such as the NMDID (whole body CT scanner), including to consider indicators of bone quality or cortical bone thickness to the geometrical information.

Concerning the modeling of the whole body external shape, several statistical models exist, such as the one developed at LBMC for 2 years in partnership with an industrial company. This type of model is typically based on a limited number of postures (e.g. UMTRI Humanshape). While these models can be articulated using numerical methods developed for the computer graphics industry (i.e. skinning), the result is not necessarily anatomically relevant. And because these models do not include realistic skeletal shapes, it is not possible to use detailed skeletal information to adapt the outer skin (acquired in a reference posture) to a new posture.

The assembly of internal and external models is also complex (besides the supine position). Indeed, the ability to observe simultaneously in three dimensions internal and external structures without health effects is essentially limited to magnetic resonance imaging. This is generally done in the supine position, and if the so-called open scanners allow a more flexibility (e.g. Wang et al., 2021 for a seated posture), their use remains complex in particular for large anatomical regions (Beillas et al., 2009). Relationships between external and internal structures exist for some regions or joints (cf. Peng et al, 2015; Nérot et al., 2016). However, joining these statistical models into a full body would require validation work to ensure the relevance of the combination. The change from supine to a given position (e.g. seated) is also of particular interest as data are available for the supine position.

Objectives

The main objective of this thesis will be to develop a first whole body parametric model including both internal and external structures, fully articulated and with statistical shape models of the components. The work will build on previous lab efforts (data, methods, etc.), open databases, and if needed targeted data collections. The model will be published under an open license to promote its improvement and various applications in biomechanics. It will thus be able to serve as a structuring research platform both in the laboratory and outside.

The scientific issues that will be addressed during the thesis will include the preservation of coherence between internal and external shapes during repositioning, the assembly of data from different sources or postures, and the validation needs of such models.

During the thesis, the model will be connected to the PIPER open source framework and models in order to verify its performance and demonstrate its use in seating comfort and safety applications.

Approach

After a review of existing models, data and methods, the following phases are envisioned:
* Complete and assemble statistical shape models of the skeleton and outer skin. This will build on work already done in the laboratory.
* Articulation of the skeleton models (integrating the simulation of contacts) and the outer skin (with skinning methods from the literature).
* Assembly by constraints and external/internal relationships, then optimization. These internal & external constraints can be enriched by collecting new data. These data can also be used for validation.
* Refinement of the skeleton and skin by kinematic methods (observation of the movement and integration of new constraints on the skeleton).
* Validation from experimental data.
* Applications (interactions with models available in the laboratory and applied to comfort, safety or health).

Profile
The candidate will have a background in Mechanics or Applied Computer Science, with a strong interest in biomechanics and modeling.

References
Beillas P, Lafon Y, Smith FW. (2009) The effects of posture and subject-to-subject variations on the position, shape and volume of abdominal and thoracic organs. Stapp Car Crash J. Nov;53:127-54.

Grébonval, C., Trosseille, X., Petit, P., Wang, X., and Beillas, P. (2021). Effects of seat pan and pelvis angles on the occupant response in a reclined position during a frontal crash. PLOS ONE 16.

Nerot, A., W. Skalli, and X. Wang. (2016) “A Principal Component Analysis of the Relationship between the External Body Shape and Internal Skeleton for the Upper Body.” Journal of Biomechanics 49, no. 14 (October 3, 2016): 3415–22. .

Peng J., Panda J., Van Sint Jan S., Wang X. (2015) Methods for determining hip and lumbosacral joint centers in a seated position from external anatomical landmarks. Journal of Biomechanics 48 (2015), 396-400

Savonnet L, Duprey S, Van Sint Jan S, Wang X (2019) Pelvis and femur shape prediction using principal component analysis for body model on seat comfort assessment. Impact on the prediction of the used palpable anatomical landmarks as predictors. PLoS ONE 14(8):e0221201.

Wang X., Savonnet L., Theodorakos I., Beurier G., Duprey S.(2019) Biomechanical human models for seating discomfort assessment. DHM and posturography, Editors S. Scataglini and G. Paul. Academic Press, 643-659.

Wang X., Savonnet L., Capbern L & Duprey S. (2021) A Case Study on the Effects of Foam and Seat Pan Inclination on the Deformation of Seated Buttocks Using MRI, IISE Transactions on Occupational Ergonomics and Human Factors

Keywords : Statistical shape modelling, human, skeleton, skin enveloppe, posture, safety, comfort, health, open source

Main host Laboratory - Referent Advisor	TS2 - LBMC - BEILLAS Philippe tél. : +33 472142371
Director of the main host Laboratory	MITTON David -
PhD Speciality	Mécanique/Biomécanique
Axis of the performance contract	1 - COP2017 - Efficient transport and safe travel
Main location	Bron
Doctoral affiliation	UNIVERSITE CLAUDE-BERNARD-LYON 1
PhD school	MEGA (MECANIQUE, ENERGETIQUE, GENIE CIVIL, ACOUSTIQUE)
Planned PhD supervisor	WANG Xuguang - Université Gustave Eiffel - TS2 - LBMC
Planned PhD co-supervisor	LAFON-JALBY Yoann - -
Planned financing	Contrat doctoral - Université Gustave Eiffel
Abstract Context: In the transportation field, vehicle automation requires solutions that provide information on occupant status, posture, activities, and more generally, any biomechanical characteristic that would be relevant to assess seating comfort and injury risk in case of accident. This potentially includes both internal and external information. It should be possible to generate this information with very partial sensor data (monitoring concept). Before that, the same type of data should be used to design new interior concepts and layouts, for both comfort and safety. For this, customizable/parametric models of the human body using deformable finite element and rigid multi-body methods are needed to assess seating comfort (Wang et al., 2019) as well as safety (Grébonval et al., 2021). In health, a large amount of imaging data is available today and has enabled the development of segmentation algorithms based on learning and statistical shape models of bones. However, these data are typically obtained in the supine position and do not include the whole body. For example, prediction of spinal curvature from the outer shape is an active field of research (Nérot et al., 2016). In sports and clothing, this type of model combining a statistical articulated envelope with a model of anatomical internal joints is helpful in design for sports and leisure retail companies (e.g. ongoing project with a sports equipment manufacturer). In any case, if external and internal models exist, there is no statistical shape model of the of the whole body that includes external and internal structures and that can be realistically positioned. Such a model would have many applications in predicting most likely postures or internal characteristics from partial data (external patient measurements, real-time data in a vehicle for comfort or safety, markerless measurements, etc.). However, the development of such a model presents a number of challenges and gaps. For the skeleton, existing results are partial. Research efforts are focused on the creation of local statistical shape models (femur, knee joint, etc.). The LBMC contributes to this effort with the creation of statistical shape models (pelvis and femur, Savonnet et al., 2019, but also thorax, lower limb, skull). But the assembly of these parts from different populations/data sources remains a scientific challenge to ensure consistency and relative positioning of anatomical structures. A strong potential exists however with open databases such as the NMDID (whole body CT scanner), including to consider indicators of bone quality or cortical bone thickness to the geometrical information. Concerning the modeling of the whole body external shape, several statistical models exist, such as the one developed at LBMC for 2 years in partnership with an industrial company. This type of model is typically based on a limited number of postures (e.g. UMTRI Humanshape). While these models can be articulated using numerical methods developed for the computer graphics industry (i.e. skinning), the result is not necessarily anatomically relevant. And because these models do not include realistic skeletal shapes, it is not possible to use detailed skeletal information to adapt the outer skin (acquired in a reference posture) to a new posture. The assembly of internal and external models is also complex (besides the supine position). Indeed, the ability to observe simultaneously in three dimensions internal and external structures without health effects is essentially limited to magnetic resonance imaging. This is generally done in the supine position, and if the so-called open scanners allow a more flexibility (e.g. Wang et al., 2021 for a seated posture), their use remains complex in particular for large anatomical regions (Beillas et al., 2009). Relationships between external and internal structures exist for some regions or joints (cf. Peng et al, 2015; Nérot et al., 2016). However, joining these statistical models into a full body would require validation work to ensure the relevance of the combination. The change from supine to a given position (e.g. seated) is also of particular interest as data are available for the supine position. Objectives The main objective of this thesis will be to develop a first whole body parametric model including both internal and external structures, fully articulated and with statistical shape models of the components. The work will build on previous lab efforts (data, methods, etc.), open databases, and if needed targeted data collections. The model will be published under an open license to promote its improvement and various applications in biomechanics. It will thus be able to serve as a structuring research platform both in the laboratory and outside. The scientific issues that will be addressed during the thesis will include the preservation of coherence between internal and external shapes during repositioning, the assembly of data from different sources or postures, and the validation needs of such models. During the thesis, the model will be connected to the PIPER open source framework and models in order to verify its performance and demonstrate its use in seating comfort and safety applications. Approach After a review of existing models, data and methods, the following phases are envisioned: * Complete and assemble statistical shape models of the skeleton and outer skin. This will build on work already done in the laboratory. * Articulation of the skeleton models (integrating the simulation of contacts) and the outer skin (with skinning methods from the literature). * Assembly by constraints and external/internal relationships, then optimization. These internal & external constraints can be enriched by collecting new data. These data can also be used for validation. * Refinement of the skeleton and skin by kinematic methods (observation of the movement and integration of new constraints on the skeleton). * Validation from experimental data. * Applications (interactions with models available in the laboratory and applied to comfort, safety or health). Profile The candidate will have a background in Mechanics or Applied Computer Science, with a strong interest in biomechanics and modeling. References Beillas P, Lafon Y, Smith FW. (2009) The effects of posture and subject-to-subject variations on the position, shape and volume of abdominal and thoracic organs. Stapp Car Crash J. Nov;53:127-54. Grébonval, C., Trosseille, X., Petit, P., Wang, X., and Beillas, P. (2021). Effects of seat pan and pelvis angles on the occupant response in a reclined position during a frontal crash. PLOS ONE 16. Nerot, A., W. Skalli, and X. Wang. (2016) “A Principal Component Analysis of the Relationship between the External Body Shape and Internal Skeleton for the Upper Body.” Journal of Biomechanics 49, no. 14 (October 3, 2016): 3415–22. . Peng J., Panda J., Van Sint Jan S., Wang X. (2015) Methods for determining hip and lumbosacral joint centers in a seated position from external anatomical landmarks. Journal of Biomechanics 48 (2015), 396-400 Savonnet L, Duprey S, Van Sint Jan S, Wang X (2019) Pelvis and femur shape prediction using principal component analysis for body model on seat comfort assessment. Impact on the prediction of the used palpable anatomical landmarks as predictors. PLoS ONE 14(8):e0221201. Wang X., Savonnet L., Theodorakos I., Beurier G., Duprey S.(2019) Biomechanical human models for seating discomfort assessment. DHM and posturography, Editors S. Scataglini and G. Paul. Academic Press, 643-659. Wang X., Savonnet L., Capbern L & Duprey S. (2021) A Case Study on the Effects of Foam and Seat Pan Inclination on the Deformation of Seated Buttocks Using MRI, IISE Transactions on Occupational Ergonomics and Human Factors
Keywords :	Statistical shape modelling, human, skeleton, skin enveloppe, posture, safety, comfort, health, open source

List of topics

Applications closed