The MICROCARD project: Advancing Cardiac Electrophysiology with Exascale Simulation

After more than three years since its start in April 2021, the MICROCARD project has reached its conclusion. This ambitious EuroHPC JU-funded project aimed at revolutionising cardiac electrophysiology simulations by creating an exascale application platform for cell-by-cell simulations.

The platform has been developed through collaboration between HPC experts, numerical scientists, biomedical engineers, and biomedical scientists from both academia and industry. The MICROCARD project involved ten partners from six different countries: France, Norway, Italy, Switzerland, Germany, and Austria.

With an overall budget of €5,800,000 from Horizon Europe, the MICROCARD project represents a significant effort in advancing biomedical simulations. 

In this interview, the EuroHPC JU spoke with Mark Potse, the coordinator of the MICROCARD project, to discuss its achievements and future implications.

Can you please describe the MICROCARD project in your own words?

The MICROCARD project aimed to develop a software that can simulate the human heart with micrometre resolution. This advanced software will enable researchers to investigate how cardiac arrhythmias develop in structurally abnormal tissues, which are often involved in these serious conditions. Achieving this level of detail requires a million times more computational power than current models can provide, necessitating the use of exascale computers or beyond.

What have been the key objectives for the MICROCARD project and what progress has been made?

Our main objectives were to develop both the software needed to simulate the human heart and the tools required to create the detailed meshes for these simulations. These meshes involve constructing data structures that describe the heart tissue's geometry with billions or even trillions of tiny elements. 

For the simulation software, we developed and applied sophisticated mathematical techniques to accurately solve the equations that model the heart’s behavior. These methods are essential for capturing the complex dynamics of the heart and ensuring that the simulations provide precise and meaningful results.

We also created specialised tools to handle complex calculations more effectively. Additionally, we developed a dedicated compiler that translates the simulation’s equations into optimised code for various types of supercomputing hardware, such as GPUs and vector CPUs.

Lastly, we have also enhanced existing meshing software to handle not just the enormous size of our meshes but also the exceptional complexity of biological tissue.

Can you give some concrete examples of how your project supports European HPC users and how it promotes greener and more sustainable supercomputing?

We are enhancing our simulation software by building on the existing OpenCARP code and integrating our improvements. These upgrades make the software much more efficient for users: in some cases, it runs up to 20 times faster than before, and it also performs significantly better on GPUs. Additionally, we have improved how the software handles data output, making it more efficient even during complex simulations.

These enhancements are now available to a broad range of users, including those working with less detailed models on HPC systems. We also strengthened the Mmgtools remeshing software, a software suite used for generating and refining meshes, which are essential for creating detailed models in simulations.

These improvements will benefit cardiologists and their patients by providing more accurate and detailed heart simulations, which can aid in diagnosing and treating heart conditions.

What were the main challenges you encountered during the project's development, if any?

Perhaps the most daunting task we have faced, was finding people who could code advanced mathematics in HPC software. This combination of skills is rare, and our competitors have substantial financial resources. 

Also, communicating between all the different disciplines involved in this project was not always easy, but I believe we managed very well.

How is the development of such a project supporting the ambition of the EuroHPC JU to make Europe a world-leader in supercomputing?

Using world-class supercomputers has become a very challenging task due to the scale and complexity of current machines. Projects like ours are crucial for helping user communities learn how to effectively utilise these technologies and for HPC experts to gain insight into the needs and requirements of users. While we learn, we are also educating the next generation of HPC programmers, and we are informing a larger audience about the challenges and benefits of using supercomputers.

What’s next for your project and results developed under this project?

We are extremely happy that EuroHPC has decided to fund a Centre of Excellence that will follow up on the MICROCARD project. This will give us the opportunity to test and deploy our software on actual exascale supercomputers, which are now being procured by the EuroHPC JU. 

We will use the lessons we learned in MICROCARD to improve our approach: for example, we will concentrate on a single discretisation scheme, which means we will concentrate our efforts on using one specific method to divide and represent continuous data or equations into discrete, manageable elements for simulation and analysis. 

We will also invest more effort in the production of geometric models – a topic that we underestimated when we devised the MICROCARD project. 

Lastly, we are also thinking about future projects to give these models a solid foundation in imaging data, and of course, research projects that will use our software. An exascale code is like a spaceship: you don’t just give it away to the public to play with it; it takes a large organisation to deploy it in a useful way.