Lattice-Boltzmann simulations of cerebral blood flow

Computational haemodynamics play a central role in the understanding of blood behaviour in the cerebral vasculature, increasing our knowledge in the onset of vascular diseases and their progression, improving diagnosis and ultimately providing better patient prognosis. Computer simulations hold the potential of accurately characterising motion of blood and its interaction with the vessel wall, providing the capability to assess surgical treatments with no danger to the patient. These aspects considerably contribute to better understand of blood circulation processes as well as to augment pre-treatment planning. Existing software environments for treatment planning consist of several stages, each requiring significant user interaction and processing time, significantly limiting their use in clinical scenarios. The aim of this PhD is to provide clinicians and researchers with a tool to aid in the understanding of human cerebral haemodynamics. This tool employs a high performance fluid solver based on the lattice-Boltzmann method (coined HemeLB), high performance distributed computing and grid computing, and various advanced software applications useful to efficiently set up and run patient-specific simulations. A graphical tool is used to segment the vasculature from patient-specific CT or MR data and configure boundary conditions with ease, creating models of the vasculature in real time. Blood flow visualisation is done in real time using in situ rendering techniques implemented within the parallel fluid solver and aided by steering capabilities; these programming strategies allows the clinician to interactively display the simulation results on a local workstation. A separate software application is used to numerically compare simulation results carried out at different spatial resolutions, providing a strategy to approach numerical validation. This developed software and supporting computational infrastructure was used to study various patient-specific intracranial aneurysms with the collaborating interventionalists at the National Hospital for Neurology and Neuroscience (London), using three-dimensional rotational angiography data to define the patient-specific vasculature. Blood flow motion was depicted in detail by the visualisation capabilities, clearly showing vortex fluid ow features and stress distribution at the inner surface of the aneurysms and their surrounding vasculature. These investigations permitted the clinicians to rapidly assess the risk associated with the growth and rupture of each aneurysm. The ultimate goal of this work is to aid clinical practice with an efficient easy-to-use toolkit for real-time decision support