Improving the modelling of cardiac arrhythmia
Oxford e-Research Centre (OeRC)
EPCC, University of Edinburgh
150ms of cardiac monodomain activity showing a wave of electrical activation propagating over the heart. The mesh contains around 153 millions elements and 26 million nodes. Data and simulations courtesy of S.A Niederer, funded by EPSRC Grand Challenge grant EP/F059361/1 and an EPSRC Life Sciences Interface Fellowship.
Left: Simulation of a complex arrhythmia episode within a highly detailed computational model of the
rabbit ventricles performed using the CARP simulation software.
Right: Computational finite element model of the rabbit ventricles derived directly from high resolution
MR data containing a wealth of fine-scale anatomical complexity.
Experimental MR data courtesy of the teams of Dr Peter Kohl, Dr Jurgen Schneider. Simulations performed
by Dr Martin Bishop and Dr Gernot Plank.
EPCC, in collaboration with Dr Gernot Plank of the Oxford e-Research Centre (OeRC), has been awarded funding to optimise heart-modelling software. If successful, this work will enable much greater integration of computer simulation with the operating theatre and could, ultimately, lead to personalised medicine.
Medical science is increasingly turning to computational models to study the possible effects of drugs and surgical interventions, before moving on to patient trials. One active area of research is in heart modelling. The structure of a patient's heart can be obtained through MRI scans, this data is then placed on a computer and used to construct a model heart. Researchers can study the electrical activity in this model heart, which controls heart beats. Problems such as arrhythmia can be identified and possible surgical interventions can be tested on the model before being used on the patient.
To obtain accurate guiding information, these models must be extremely detailed, requiring the use of large HPC systems. The current state-of-the-art software requires a number of hours on such systems to simulate a single beat of a human heart. Computational studies of arrhythmias require many hundreds of beats and so the current performance is not yet good enough to provide useful surgical data.
The most efficient software package available for heart modelling is the Cardiac Arrhythmia Research Package (CARP). Recently, EPCC has been working with Dr Gernot Plank and the OeRC, lead developers of CARP, to identify performance problems in the code and suggest how they should be addressed. A two-month study, conducted as part of DEISA Extreme Computing Initiative project muHeart, identified three main areas of interest. Preliminary work addressing these areas suggests that eliminating the identified performance issues could reduce the time required for heart beat simulations from hours to minutes. This is an exciting development, since it brings the prospect of using computational models as part of the surgical workflow very much closer.
Dr Plank said: "The profiling work EPCC has carried out has been extremely important in guiding plans for future model development and I am very much looking forward to continuing the work to extend these results to performance improvements. The performance improvements will enable us to simulate the electrical activity of the heart in health and under pathological conditions at an unprecedented level of anatomical and functional detail which will, eventually, pave the way to personalised medicine."
The funding to carry out the performance optimisation was awarded to EPCC and Dr Gernot Plank through the dCSE programme. The work will be carried out on HECToR during 2010.
See also:
- G. Plank et al.,"Generation of histo-anatomically representative models of the individual heart: tools and application," Phil Trans Roc Soc, 2009.
- DEISA Extreme Computing Initiative