Cardiac Modeling - Part 2       

Medical, Health Informatics and Computational Biology Accomplishment | 2012 - 2015

IBM researchers: Jeremy Rice, Slava Gurev, Matthias Reumann

Collaborators: University of Rochester, Lawrence Livermore National Laboratory, U.S. Food and Drug Administration

Where the work was done: IBM T.J. Watson Research Center

What we accomplishedFollowing up on original cardiac modeling work from 2008-2011, IBM has continued to improve the state of the art in scientific content as well as enabling new classes of models by improving algorithms and application of HPC platforms.

Related links:

  • Combination in silico and in vitro methods to predict risk of arrhythmia in Long QT 1 patients
  • High resolution models of the electrophysiology of the heart were  run at unprecedented speeds on Sequoia, a 96-rack Blue Gene Q installation at LLNL.
    • Richards, DF, Glosli, JN, Draeger, EW, Mirin, AA, Chan, B, Fattebert, J-L, Krauss, W, Oppelstrup, T, Bulter, CJ, Gunnels, JJ, Gurev, V, Kim, C, Magerlein, J, Reumann, M, Wen, HF, and Rice, JJ. Toward Real-Time Simulation of Cardiac Electrophysiology in a Human Heart at High Resolution. Computer Methods in Biomechanics and Biomedical Engineering 16(7):802-805 (2013).
    • Shows power of HPC applied to biology - results were roughly 1200 times faster than previously reported - was finalist in Gordon Bell Prize
    • The ability to work at high resolution and with high throughput enabled new kind of scientific research in re-entrant arrhythmia.  We are the first group to be able to simulate Torsade de Pointes in high detail.  These is an important kind of arrhythmia that can be brought on buy drug, disease or genetic variants in the population. While risk is low for an individual, when drugs are given in millions of dose a year, this arrhythmia cause substantial numbers of death.  Trying to quantify these risk levels is a challenge to the scientific community and regulatory agencies.
  • High resolution model of mechanics sets new standard in the field for the level of anatomic detail.
    • Gurev, V*, Pathmanathan, P*, Fattebert, J-L, Wen, HF, Magerlein, J, Gray, RA, Richards, DF and Rice, JJ. A high-resolution computational model of the deforming human heart. Biomechanics & Modeling in Mechanobiology (online Jan. 1, 2015)
    • This work required the development of new algorithms that were stable and accurate when simulating heart mechanics in very high resolution.  Typical models in the field were limited to a few hundred finite elements whereas the new approach has been demonstrated to work on over 1 million elements.
    • The heart is a complex anatomical structure that could only be roughly approximated with early approaches. This new approach enables the representation of trabeculae and papillary muscle in the heart as well as infarcts, which are injuries that can be highly localized with very fine structures that could not be directly represented with earlier coarser methods.  These capabilities will all the normal and pathological states of the heart to be simulated with more realism, and the new approach will allow for highly personalized models of the heart to guide treatment on an individual basis.

Image credit: IBM Research