A deeper understanding of the fluid dynamics in biomedical geometries can be a valuable tool in the study of the human circulatory system and in the improvement of prosthetic devices such as coronary artery bypass grafts (CABG). Clinical studies and experience suggest that bypass grafts tend to fail after some years due to the process of restenosis, whereas computational fluid dynamics can help to assess unfavorable flow conditions near the anastomosis of the CABG, which can be strictly related to the onset of restenosis.
The aim of this talk is to describe a new, combined framework for the simulation of blood flows in three-dimensional patient-specific CABGs. This framework is based on the combination of several ingredients, ranging from medical images, suitable computational reduction techniques for efficient numerical blood flow simulations and shape representation techniques.
More in details:
1. Imaging: a computational mesh is recovered from computed tomography scans of a few selected patient-specific geometries, including the aorta, the coronary arteries and (multiple) bypass grafts. The current clinical trial aims at the investigation of representative cases of different grafting procedures, grafting materials and native coronary disease.
2. Computational reduction for patient-specific geometries: performing real-time computational studies on patient-specific geometries is a crucial requirement to cut down large computational costs arising from complex optimization problems. Model order reduction techniques, based in particular on proper orthogonal decomposition and the reduced basis method, enable to evaluate blood flows for different flow conditions (such as Reynolds/Womersley numbers and aortic inflow boundary conditions) in a rapid and reliable way.
3. Geometrical parametrization and reduction for representative surrogate geometries: in order to perform local changes to the reconstructed geometry in a flexible and reliable way, we exploit suitable shape parametrization techniques. Once surrogate geometries have been extracted from the medical images to describe a set of representative configurations, a low dimensional parametrization is employed to perform local changes to the geometry of the grafts. Volume based parametrizations (e.g. deformation based on free-form deformations or radial basis function interpolation) and centerline based parametrizations may be considered to this goal.
Some results for a few representative patient-specific geometries are going to be discussed during the talk. The collaboration with Dr. Roberto Scrofani, M.D. (division of Cardiac Surgery, Ospedale L. Sacco, Milano) and Dr. Sonia Ippolito, M.D. (division of Radiology, Ospedale L. Sacco, Milano) is gratefully acknowledged.