An integrated geometric modelling framework for patient-specific computational haemodynamic study on wide-ranged vascular network

Patient-specific haemodynamic computations have been used as an effective tool in researches on cardiovascular disease associated with haemodynamics such as atherosclerosis and aneurysm. Recent development of computer resource has enabled 3D haemodynamic computations in wide-spread arterial network but there are still difficulties in modelling vascular geometry because of noise and limited resolution in medical images. In this paper, an integrated framework to model an arterial network tree for patient-specific computational haemodynamic study is developed. With this framework, 3D vascular geometry reconstruction of an arterial network and quantification of its geometric feature are aimed. The combination of 3D haemodynamic computation and vascular morphology quantification helps better understand the relationship between vascular morphology and haemodynamic force behind ‘geometric risk factor’ for cardiovascular diseases. The proposed method is applied to an intracranial arterial network to demonstrate its accuracy and effectiveness. The results are compared with the marching-cubes (MC) method. The comparison shows that the present modelling method can reconstruct a wide-ranged vascular network anatomically more accurate than the MC method, particularly in peripheral circulation where the image resolution is low in comparison to the vessel diameter, because of the recognition of an arterial network connectivity based on its centreline.