The study of blood flow in human arterial bifurcations by the combination of CFD and MRI
Atherosclero8is represents a major health problem in the western world. The local haemodynamics is believed to be an initiating and localizing factor in this multilactorial disease process. To fully understand this interaction it is important to obtain detailed information about the local haemodynamics in accurate models of the hnrnan vascular system. Because of the complexity of arterial geometry, in mvo velocity measurements are subject to large errors by currently available techniques. It is also difficult to construct the highly irregular arterial bifurcation model for in vitro investigations. By using a combination of two new methodologies,namely magnetic resonance angiography (MRA) and computational fluid dynamics CFD), the precise patterns of flow anticipating the onset of disease at arterial bifurcations can now, in principle, be determined. However, flow simulations based on in vzvo data directly acquired from clinical measurements have rarely been performed, due to difficulties involved in converting medical images into a data set that CFD software packages can accept. ,In this study, a computer modelling technique, which integrates dinically acquired MR angiograms, image processing and CFD, for the reconstruction of 3D blood flow patterns in realistic arterial geometry, was developed. In the procedure, human arteries are scanned non-invasively by MR angiography. With the MR angiograms, image processing and 3D reconstruction are performed and structured numerical grid is generated for the arteries scanned. Together with MR in tnvo measured velocity profiles at the boundary planes of the model, CFD simulations are undertaken. To test the capability and reliability of the whole procedure, two examples are given, of the human abdominal and right carotid bifurcations. The complete haemodynamic patterns obtained allow a full clinical understanding to be gained of individual patient behaviour. Aspects such as wall shear stress variation, secondary flow and flow separations are demonstrated. The problem of quantitative reliability of the predictions is discussed in some depth.