Periodic Reporting for period 2 - BREATHE (Unlocking vital mysteries in respiratory biomechanics)
Reporting period: 2023-03-01 to 2024-08-31
While the human lung is undoubtedly an essential organ, and respiratory diseases are leading causes of death and disability in the world, there still exist a lot of mysteries wrt vital processes. The main reason for this is the complete lack of measurement methods or medical imaging techniques that would allow to study dynamic processes in essential parts of a living human lung. While this would be a perfect setup for computational modeling, existing models suffer from severe constraints disabling them to unveil those essential secrets. This project aims to build on a number of most promising recent advances in modeling and high-performance simulation to present the first comprehensive computational model of the respiratory system. For this purpose, it builds upon a recent exascale-ready incompressible flow solver, toughen it up for lung specific challenges and enrich it with multiphysics capabilities to capture tissue interaction and gas transport. Parts of the respiratory zone will be represented by multiphase poroelastic media and novel pleural boundary conditions will be developed. The coupled pulmonary circulation will be included and represented by an embedded reduced dimensional network and additional phases. In order to appropriately individualize the model and also being able to adapt it during disease progression, a novel physics-based probabilistic learning approach will be developed. This will allow to use most of the very diverse and scarce data in clinical settings. Finally, special models will be developed to bridge to the micro scale. The models developed and studied here will provide unprecedented insights for biomedical scientists, and practitioners at the same time, and will help to substantially reduce elaborate animal and multicenter studies. This will be a crucial step in order to establish a shift of paradigm in health care. Novel models/tools developed here will also be very useful in other areas of biomedical engineering and beyond.
The main work related to the individual work packages is listed below in short form:
WP1:
- Both H-DIV conforming elements as well as simplex elements have been made available within the HPC CFD code
- The new concept (formulation and implementation) of simplex elements for our HPC CFD code could be so much further improved that it is now close to the range of the best-performing hexahedral (tensor product) elements.
- A novel concept has been developed that allows the use of matrix-free CFD solver together with matrix-free as well as matrix-based structural solver not only for partitioned fluid-structure interaction solver but also for the desired monolithic solver. The high-performance implementation of this general concept is in its' final stage. The implementation is currently exemplified for Poisson equations on both sides, but already tackles the complexity that allows for incorporation of block solver and multigrid (algebraic and geometric) techniques for both coupled fields.
WP2:
- Development and implementation of two novel poroelastic approaches for lungs (one for a two-phase model based on velocity and pressure and a general multiphase and multispecies model purely pressure based)
- Formulation for highly anisotropic and inhomogeneous (to capture effects of airway trees) poroelasticity
- Development and implementation of a new hybrid multi-dimensional coupled multiphase (and multispecies) approach has been achieved. This is the first model of its kind that allows for the highly relevant coupling of the respiratory system with pulmonary circulation. The essentials of the model are a homogenized multiphase multispecies background domain and discrete coupled with discrete airways and blood vessels.
- Another formulation for highly anisotropic and inhomogeneous poroelasticity has been developed (to capture the effects of airway tree structures in homogenized domains)
- We could also improve our virtual Electro Impedance Tomography model and develop a new pipeline in order to allow for validation via patient studies
- A novel alveolar recruitment/derecruitment model has been developed
WP3:
- Several improved approaches for inverse analysis have been developed
- Important steps have been realized for adapting such approaches to special multiphysics models and investigated on test examples
- Some first combinations of these new approaches have been employed to automatically identify patient-specific model parameters for reduced dimensional models of the respiratory system
WP4:
- First steps have been realized to model collagen fibers and considering intermolecular effects
- A careful investigation of numerical problems caused by jumps of forces at the end of fibrils has been performed and some first remedies have been introduced to remedy the associated convergence issues
WP1:
- Both H-DIV conforming elements as well as simplex elements have been made available within the HPC CFD code
- The new concept (formulation and implementation) of simplex elements for our HPC CFD code could be so much further improved that it is now close to the range of the best-performing hexahedral (tensor product) elements.
- A novel concept has been developed that allows the use of matrix-free CFD solver together with matrix-free as well as matrix-based structural solver not only for partitioned fluid-structure interaction solver but also for the desired monolithic solver. The high-performance implementation of this general concept is in its' final stage. The implementation is currently exemplified for Poisson equations on both sides, but already tackles the complexity that allows for incorporation of block solver and multigrid (algebraic and geometric) techniques for both coupled fields.
WP2:
- Development and implementation of two novel poroelastic approaches for lungs (one for a two-phase model based on velocity and pressure and a general multiphase and multispecies model purely pressure based)
- Formulation for highly anisotropic and inhomogeneous (to capture effects of airway trees) poroelasticity
- Development and implementation of a new hybrid multi-dimensional coupled multiphase (and multispecies) approach has been achieved. This is the first model of its kind that allows for the highly relevant coupling of the respiratory system with pulmonary circulation. The essentials of the model are a homogenized multiphase multispecies background domain and discrete coupled with discrete airways and blood vessels.
- Another formulation for highly anisotropic and inhomogeneous poroelasticity has been developed (to capture the effects of airway tree structures in homogenized domains)
- We could also improve our virtual Electro Impedance Tomography model and develop a new pipeline in order to allow for validation via patient studies
- A novel alveolar recruitment/derecruitment model has been developed
WP3:
- Several improved approaches for inverse analysis have been developed
- Important steps have been realized for adapting such approaches to special multiphysics models and investigated on test examples
- Some first combinations of these new approaches have been employed to automatically identify patient-specific model parameters for reduced dimensional models of the respiratory system
WP4:
- First steps have been realized to model collagen fibers and considering intermolecular effects
- A careful investigation of numerical problems caused by jumps of forces at the end of fibrils has been performed and some first remedies have been introduced to remedy the associated convergence issues
For progress see above and the expected results are still as ambitious as in the original proposal.