B-BeST: Bringing Biomechanical Simulations to clinical Target
The goal of this project is to test computational fluid dynamics simulations of the thoracic aorta on a HPC infrastructure and give a step forward towards bringing current developments into medical practise. With previous CINECA projects, we have demonstrated how simulations from a single patient can help medical decision making. However, there are still two limitations on taking computational simulations to the bedside.
On the one hand, timescales of medical practice and CFD are mismatched: while doctors need answers in minutes or hours to make decisions, simulations take days or weeks to be performed. On the other hand, while single patient studies have their merits, large data-set are necessary for a strong impact in medical research and therefor computational time is also here of the essence.
The goal of this project is to exploit HPC capabilities on the new cluster MARCONI in order to:
- port the code to the new Broadwell and KNL infrastructures;
- considerably reduce the computational time of blood flow simulations within the thoracic aorta thus assessing the feasibility of taking CFD to the bedside;
- simulate a big data-set of undiseased thoracic aortic geometries with CFD in order to better assess the impact of identified parameters on thoracic aortic blood flow.
To the date, many groups have been working on aortic CFD simulations, both on the thoracic and the abdominal aorta. Within our group at UniPV in collaboration with IRCCS Policlinico San Donato, we are building two databases within the iCardioCloud project. The first one consists on CFD simulations of aneurysmatic aortas both before and after TEVAR and virtually constructed geometries simulating the procedure in-silico. The otherone consists on healthy aortas with different types of arch geometry in order to quantify the impact of the arch shape in the positional stability of the endograft.
Through the LISA 2016-2018 initiative, the CompMech Group @ UniPV has been awarded a total of 1.720.320 core hours to work on this project.