Personalised Surgery

Patient specific simulation for surgical planning

Predictive modelling of haemodynamics

Cardiovascular disease is the cause of a large number of deaths in the developed world. Cerebral blood flow behaviour plays a crucial role in the understanding, diagnosis and treatment of the disease; problems often arise due to anomalous blood flow behaviour in the neighbourhood of vascular bifurcations and aneurysms within the brain, leading to stroke for example, although the details are not yet well understood. Experimental studies are often impractical owing to the difficulty of measuring behaviour in human subjects; however, computer tomography (CT) and three-dimensional rotational angiography (3DRA) enable static and dynamical data acquisition.

In combination with these image acquisition methods, modelling and simulation have a crucial role to play in haemodynamics. Simulation offers the clinician the possibility of performing non-invasive, patient specific, virtual interventions to plan and study the effects of proposed courses of surgical treatment—with no danger to the patient—including support for diagnosis, therapy and planning of vascular treatment. Simulation techniques also offer the prospect of modelling the poorly understood flow patterns in the normal brain and within neurovascular pathologies such as aneurysms and arterio-venous malformations (AVMs); and, indeed, to study haemodynamics in the entire human arterial tree.

Our modelling solution

Simulation input data is provided by CT 3DRA scans, from which a patient specific model of the vascular structure of the brain is built using imaging software tools. We have developed the very fast HemeLB application to execute blood flow in these models, which can act in an advisory capacity to the surgeon before (and eventually we envisage also during) surgery. In addition to the fluid flow solver, HemeLB includes an in-built ray-tracing engine, giving it real time visualization capabilities, with frames of visualization rendered on the same processor cores as the simulation. A simple desktop display and computational steering client connects to the application at run time to display the visualization and allow the user to interact with it. Simulations, used to support clinical procedures, require large numbers of processor cores, and have to be run on demand when required by the clinician.

On-time support for surgical planning

Using the e-infrastructure we have put in place to run the HemeLB application, end users can reserve time on a suitably powerful computer in advance, allowing them to run the application in real time, as and when needed, rather than when it arrives at the top of a job queue. This means that patient specific simulations using HemeLB have the potential to be scheduled into clinical workflows, and used as an aide to a surgeon planning an interventional procedure.

The figure shows a typical work flow from CT image acquisition (combining segmentation and reconstruction in a largely automated sequence of steps), to embedding the patient specific intracranial vasculature within the HemeLB flow solver, specifying the flow boundary conditions and then rendering the flow images to convey key information (on the wall shear stress) in a format suitable for clinical decision support.

 Publications

R. W. Nash, H. B. Carver, M. O. Bernabeu, J. Hetherington, D. Groen, T. Krüger and P. V. Coveney, “Choice of boundary condition for lattice Boltzmann simulation of moderate-Reynolds-number flow in complex domains”, Physical Review E, (2013), In Press.

M. Bernabeu, R. Nash, D. Groen, H. Carver, J. Hetherington, T. Krüger and P. V Coveney, “Impact of blood rheology on wall shear stress in a model of the middle cerebral artery”, J R Soc Interface Focus, 3 (2), 20120094, (2013), DOI: 10.1098/rsfs.2012.0094

D. Groen, J. Borgdorff, C. Bona-Casas, J. Hetherington, R. W. Nash, S. J. Zasada, I. Saverchenko, M. Mamonski, K. Kurowski, M. O. Bernabeu, A. G. Hoekstra and P. V. Coveney, “Flexible composition and execution of high performance, high fidelity multiscale biomedical simulations”, J R Soc Interface Focus, 3 (2), 20120087, (2013), DOI: 10.1098/rsfs.2012.0087