An interview with Prof. Kyriakos C. Giannakoglou, Training Coordinator of the MeDiTATe project
We had a small talk with Prof. Kyriakos C. Giannakoglou, a member of the Board of the MeDiTATe project. He is a Professor at the School of Mechanical Engineering of the National Technical University of Athens (NTUA) and head of the Parallel CFD & Optimization Unit of the same School. His research interests include the development of CFD-based analysis and optimization methods and tools in fluid mechanics, incl. adjoint & evolutionary algorithms, extended to multi-disciplinary applications (aerostructural, aerothermal, aeroacoustic). His research group has developed the publicly available continuous adjoint for shape optimization in the OpenFOAM environment. In MeDiTATe, he is supporting/training ESR 04, ESR 05 and ESR 06 to extend/adapt an existing CFD software on GPUs, coupled with a structural analysis tool, to simulate aneurysm-related flows, at reduced cost, as a part of a digital twin. He has participated in numerous projects funded directly by the industry, as well as the EU, including eight MSCA ITNs. He has supervised 29 PhDs (accomplished) whereas 10 more are in progress.
- The MeDiTATe project, a research proposal funded by the European Commission in the framework of Horizon 2020 programme. Can you tell us more about NTUA and its role in the project?
Founded in 1837, the National Technical University of Athens (NTUA) is the oldest Technical University in Greece. NTUA participates into MeDiTATe through the Parallel CFD & Optimization Unit (PCOpt) of the School of Mechanical Engineering. The PCOpt/NTUA group comprises 15 research engineers developing, among other, CFD methods and software which are also ported onto HPC systems, including clusters of GPUs. For us, MeDiTATe was a great opportunity to see a portfolio of in-house tools being adapted to biomedical flows. The three ESRs associated with the PCOpt/NTUA had to use the in-house CFD code PUMA, written in C++/CUDA and accelerated on NVIDIA GPUs; next to PUMA, the three ESRs had also the chance to familiarize themselves with the open-access CFD software OpenFOAM; having access to two codes makes their life easier and gives them the possibility to make comparisons. Let me say a few things about the work of the three ESRs, one by one.
The objective of ESR 04 was to extend/integrate and use the CFD model to simulate the flow in aneurysms, assuming rigid walls, Newtonian and non-Newtonian fluids and variable viscosity. Flow computations using PUMA running on GPU clusters are fast, and this is challenging in the context of a digital twin. Studies are expected to lead to estimates of the aneurysm rupture risk, occurring when the stresses acting on the arterial wall exceed its failure strength. Next to them, Deep Neural Networks (DNNs) predict the failure points of the arterial wall, the origin and the progress of the aneurysms. ESR 05 extends the same background code to cope with moving walls, by incorporating a generic wall thickness model and Fluid-Structure Interaction (FSI) techniques. The structural problem within the solid region is solved using Finite Element Method (FEM) tools. A DNN is trained based on available real data to predict the cerebral aneurysm evolution and validate existing Fluid-Solid-Growth models of cerebral aneurysm evolution. Finally, the main objective of ESR 06 is to identify hemodynamic variables which are of greatest importance in aneurysm stabilization or rupture, through sensitivity studies. Uncertainty quantification (UQ) techniques are used for quantities of interest, such as the asymmetry metric, the saccular index, etc. The sensitivity of the simulation results changes in the input variable values are investigated. Shape imperfections and wall movement are realized by CAD-free techniques and a morphing/smoothing tool that adapts meshes to new boundary shapes.
- Let’s talk about the idea behind MeDiTATe. The project received funding from the EU in the Framework of Horizon 2020 programme, therefore it is clear that the proposed approach meets the EU’s ambition to exploit scientific research to improve the treatment of the cardiovascular diseases and patient care. In particular, how can the scientific community benefit from your work?
The ultimate goal of the MeDiTATe network is the creation of Medical Digital Twins; to do so, several partners, from universities, research centres, hospitals and companies, with experience in the treatment of the cardiovascular diseases and patient care and numerical methods performing the solution of fluid problems interacting with the structural analysis of their vessels should join efforts; making these people work together was the first success of MeDiTATe. It is expected that the work done by the MeDiTATe ESRs will firstly lead to new findings in pure research and, then, will find its way in practical applications. The latter will be used to help scientists working for the treatment of cardiovascular diseases by offering state of the art tools for fast and accurate predictions, using simulations. Recall that the ultimate goal is to build a Medical Digital Twin.
- Is it really doable to adapt methods developed for different applications and use them for the MeDiTATe applications?
The answer is definitely yes! For instance, the numerical simulation of the flow in blood vessels, including the more specific aneurysm studies, requires the availability of three tools: a CFD software for the analysis of flow patterns, a software for the structural analysis of the blood vessels walls (a FEM code) and, of course, a technique to make them cooperate by exchanging data, which is the so-called Fluid-Solid Interaction (FSI) technique. The literature is full of methods and tools developed for the accurate and efficient solution of all the above, developed in quite different application domains. As an example, FSI methods for use in aircraft applications are quite mature right now. With some add-ons that are really necessary if these methods are to be used in simulations, for instance, in aorta flows’, such as the use of a non-Newtonian fluid, these methods can readily be ported in the field of biomedical flows. We should also keep in mind that the majority of software developed for other engineering applications is already adequately parallelized and runs on HPC systems including clusters of GPUs. It is obvious that the type of applications MeDiTATe is dealing with can readily profit of all previous developments and achievements. This is where the three ESRs associated with the PCOpt/NTUA contributed to.
- Professor Giannakoglou, the MeDiTATe project has almost reached the end. What are your thoughts on the achievements at this final stage of the project? Which skills do you think that the fellows will acquire at the end of this journey?
It is much easier for me to talk about the ESRs who worked with me in this project and spent 50% of their time in the PCOpt/NTUA. Having already exposed themselves in both the academic and industrial environment, they have already seen both sides; this is valuable piece of information that might help them, in the near future, to made decisions about the future steps in their career. While approaching the end of their involvement in MeDiTATe, I feel they have already learned a lot about the tools needed for these applications; they became familiar with CFD methods for the analysis of fluid flows, with FEM for the structural analysis and their interaction; they have extensively used them in an HPC environment, got results and know how to interpret them. They became ready to use these tools, but also any other similar tool they might have access to, for application regarding studies of aneurysms in an effort to build a digital twin, absolutely useful to people involved in the health sector. It is important that the three ESRs of PCOpt/NTUA were Mechanical Engineers and MeDiTATe managed to bridge the gap between the way engineers and medical scientists are facing this kind of problems.
- Any message for our readers?
If you are looking for a way to expose yourself simultaneously to the academic and non-academic sectors, if you wish to get high quality training on scientific/technical topics as well as transferable skills, if you are strongly interested in innovation and long-term employability, apply to a Marie Skłodowska-Curie Actions (MSCA) Innovative Training Network.