Heart disease is one of our biggest killers, but so far it has been impossible to understand exactly what goes wrong during irregular heart activity. Now, work presented at a workshop hosted by the European Science Foundation has demonstrated that powerful simulations calibrated and validated with advanced imaging techniques can create an accurate model of the heart in silico, providing a virtual test subject for future investigations.
Regular heart activity relies on an electrical stimulus that provokes a contraction in the heart muscle. A failure in either the electrical stimulus or the mechanical response causes cardiac arrhythmia - an irregular heart beat that can be life threatening. To understand the reasons why this sometimes go wrong, scientists need to take into account the size, structure and geometry of the heart.
The current techniques for measuring this process using physical experiments are limited. A crude measurement of the electrical stimulus is possible using electrodes placed in a heart, but the probes must be placed too far apart to measure more subtle changes across the heart’s surface. Another, more accurate method involves bathing the heart in a fluorescent die that emits light in proportion to the voltage experienced by the cells. However, this can only measure activity across the surface of the heart, and not inside, where important changes could be occurring.
Accurate computer simulations of the heart, however, could predict cardiac activity in three dimensions and in finer detail than these techniques, providing more information about the way the heart works.
‘As we have the equations and the data for all the parameters we can analyse exactly what causes a change in heart activity,’ Blanca Rodriguez, a fellow at Oxford University, UK and the scientific coordinator of the ESF workshop, told scientific-computing.com. ‘Any changes can be recorded, such as the voltage or current, or the transport of ions through ion channels.’
The simulations could help scientists to interpret known results, and they could even be used to perform virtual experiments of potential therapies at less cost than physical testing. Until now, however, limitations in the experimental techniques have also reduced the potential accuracy of these models, as it has been difficult to collect enough 3D data to calibrate and validate the models.
Researchers such as Rodriguez are now working to solve this by combining MRI and DTMRI scans with histological images to provide 3D information about the larger structures and geometry of the heart together with smaller details such as the orientation of the cardiac fibres, at resolutions of just 25μm. This data can then be fed into the models to provide a more precise and accurate basis for the different equations.
‘Registering the MRI scans with the histological images proved to be a big challenge,’ said Rodriguez. The task was so computationally intensive that it was necessary to make use of high-performance computing services, such as the UK national grid service, to run the algorithms in parallel across many different processors. The team also used HPC to run the actual simulations after this information had been integrated.
Rodriguez has already applied the simulations to examine the effects of electrical therapies, in which a current is applied to the heart to try to regulate irregular heart activity. The research was presented at the ESF Exploratory Workshop on European Heart Modelling and Supporting Technology in Oxford, UK.