504 Patient-specific Computational Biomechanics: Clinical and Industrial Transfer
Stéphane P.A. Bordas, University of Luxembourg
Mathias J. Brieu, LaMCube, Centrale Lille
Stéphane Cotin, Inria
Alejandro Frangi, University of Sheffield
Key words: Patient-specific simulation, quality control, inverse methods, uncertainty quantification, digital twins
ABSTRACT
One of the challenges of the XXIth century is to enable the impact of computational sciences on biomedical engineering to reach the grail of personalised medicine and its promise to improve patient-specific care.
Numerical modelling of medical therapies has the potential to help with diagnosis, enable the estimation of the impact of surgical interventions, optimise treatment schemes, train surgeons to hone their skills and guide them during the surgical interventions themselves. If we could model the effect of diet, medical treatments, such as drugs and surgery [1,2], patient care could be significantly enhanced.
The first requirement for this paradigm shift to be achieved is to model the geometry of the organs as well as any connective tissues. Next, the constitutive material models of these organs, which are known to vary significantly with age, gender and life style, must be selected, and their parameters identified using available data [4] or accurate sensors. Furthermore, interaction with medical devices must be considered. Finally, the resulting mathematical problems must be solved, often in time frames much shorter than what is common in engineering practice [5] and the uncertainty associated with each parameter must be quantified [6]. Such simulations lead to design in-silico clinical trials for medical devices testing or medical doctors training, plan or assist during treatments, to finally improve healthcare. Last point that will be addressed during this minisymposium, is to provide the users a clear and quantitative estimation of the patient-specific characteristics and error and uncertainty on measures. These measures would give confidence in the results, and provide clues on the impact of each assumption on accuracy. Such error and uncertainty quantification approaches also enable to decide which additional experiments or medical imaging data would decrease uncertainty in model identification and parameter estimations.
This minisymposium will therefore bring together researchers from the fluid and solid mechanics communities, working in the field of Medical Simulation. The purpose is to share experience on the way to develop reliable, and accurate modelling tools to answer to the expectation of biomedical engineering.
Particular subjects of interest include:
- Patient-specific geometric and mechanical characterization,
- in silico clinical trials
- uncertainty quantification & error estimation
- real-time simulation and inverse problems
- machine learning for constitutive model selection and adaptation
- experimental methods (imaging, sensors) and their connection to simulation & model updating
- high performance computing
REFERENCES
[1] Bui HP, Tomar S, Courtecuisse H, Cotin S, Bordas S. Real-time error control for surgical simulation. IEEE Transactions on Biomedical Engineering. 2017 Jun 1. http://orbilu.uni.lu/bitstream/10993/28624/1/tbme_Bui.pdf
[2] Bui HP, Tomar S, Courtecuisse H, Audette M, Cotin S, Bordas S. Controlling the Error on Target Motion through Real-time Mesh Adaptation: Applications to Deep Brain Stimulation. arXiv preprint arXiv:1704.07636. 2017 Apr 25. http://hdl.handle.net/10993/30937
[3] Ley C, Bordas S. What makes Data Science different? A discussion Involving Statistics2. 0 and Computational Sciences. https://publications.uni.lu/bitstream/10993/30235/1/BL17_final.pdf
[4] Cotin S, Delingette H, Ayache N. Real-time elastic deformations of soft tissues for surgery simulation. IEEE transactions on Visualization and Computer Graphics. 1999 Jan;5(1):62-73.
[5] Hauseux P, Hale J, Bordas S. Accelerating Monte Carlo estimation with derivatives of high-level finite element models, Computer Methods in Applied Mechanics and Engineering, 2016 http://orbilu.uni.lu/bitstream/10993/28618/1/phauseux-monte-carlo-revise...
ABSTRACT
One of the challenges of the XXIth century is to enable the impact of computational sciences on biomedical engineering to reach the grail of personalised medicine and its promise to improve patient-specific care.
Numerical modelling of medical therapies has the potential to help with diagnosis, enable the estimation of the impact of surgical interventions, optimise treatment schemes, train surgeons to hone their skills and guide them during the surgical interventions themselves. If we could model the effect of diet, medical treatments, such as drugs and surgery [1,2], patient care could be significantly enhanced.
The first requirement for this paradigm shift to be achieved is to model the geometry of the organs as well as any connective tissues. Next, the constitutive material models of these organs, which are known to vary significantly with age, gender and life style, must be selected, and their parameters identified using available data [4] or accurate sensors. Furthermore, interaction with medical devices must be considered. Finally, the resulting mathematical problems must be solved, often in time frames much shorter than what is common in engineering practice [5] and the uncertainty associated with each parameter must be quantified [6]. Such simulations lead to design in-silico clinical trials for medical devices testing or medical doctors training, plan or assist during treatments, to finally improve healthcare. Last point that will be addressed during this minisymposium, is to provide the users a clear and quantitative estimation of the patient-specific characteristics and error and uncertainty on measures. These measures would give confidence in the results, and provide clues on the impact of each assumption on accuracy. Such error and uncertainty quantification approaches also enable to decide which additional experiments or medical imaging data would decrease uncertainty in model identification and parameter estimations.
This minisymposium will therefore bring together researchers from the fluid and solid mechanics communities, working in the field of Medical Simulation. The purpose is to share experience on the way to develop reliable, and accurate modelling tools to answer to the expectation of biomedical engineering.
Particular subjects of interest include:
- Patient-specific geometric and mechanical characterization,
- in silico clinical trials
- uncertainty quantification & error estimation
- real-time simulation and inverse problems
- machine learning for constitutive model selection and adaptation
- experimental methods (imaging, sensors) and their connection to simulation & model updating
- high performance computing
REFERENCES
[1] Bui HP, Tomar S, Courtecuisse H, Cotin S, Bordas S. Real-time error control for surgical simulation. IEEE Transactions on Biomedical Engineering. 2017 Jun 1. http://orbilu.uni.lu/bitstream/10993/28624/1/tbme_Bui.pdf
[2] Bui HP, Tomar S, Courtecuisse H, Audette M, Cotin S, Bordas S. Controlling the Error on Target Motion through Real-time Mesh Adaptation: Applications to Deep Brain Stimulation. arXiv preprint arXiv:1704.07636. 2017 Apr 25. http://hdl.handle.net/10993/30937
[3] Ley C, Bordas S. What makes Data Science different? A discussion Involving Statistics2. 0 and Computational Sciences. https://publications.uni.lu/bitstream/10993/30235/1/BL17_final.pdf
[4] Cotin S, Delingette H, Ayache N. Real-time elastic deformations of soft tissues for surgery simulation. IEEE transactions on Visualization and Computer Graphics. 1999 Jan;5(1):62-73.
[5] Hauseux P, Hale J, Bordas S. Accelerating Monte Carlo estimation with derivatives of high-level finite element models, Computer Methods in Applied Mechanics and Engineering, 2016 http://orbilu.uni.lu/bitstream/10993/28618/1/phauseux-monte-carlo-revise...
