Computational Physiology

The computational physiology research team develops and applies advanced numerical methods and implements them on high-performance computing (HPC) architectures to answer complex questions in Physiology with a particular focus on biofluid dynamics. The research aims to understand fundamental pathophysiology of various disorders with the use of tools like computational fluid dynamics (CFD), magnetic resonance imaging (MRI), and mathematical models. The vision is to bridge the gap between clinicians, biologists and computational scientists through interdisciplinary collaborations.

  • Cerebrospinal Fluid Physiology

    The Cerebrospinal Fluid (CSF) is a colorless and odorless water-like fluid that surrounds the brain and the spinal cord. The CSF exhibits fascinating oscillatory flow patterns in the ventricular system. The detailed mechanisms behind CSF origin, reabsorption, and its movement remain elusive. The CSF further serves the function of metabolite clearance from the brain. In our research, we apply CFD to compute detailed CSF flow characteristics in subarachnoid spaces (SAS) obtained from clinical imaging. We focus on various aspects ranging from pathologies like Chiari malformation and Syringomyelia to fundamental questions about CSF pulsations.

    Velocity colored instantaneous Q-isosurfaces during peak diastole of the last cardiac cycle in the SAS of one control subject (left) and two Chiari patients (middle and right). Simulations were conducted on 96’000 cores of the HazelHen supercomputer installed in Stuttgart, Germany
  • Physiological Flow Instabilities

    Physiological flows like blood flow in the cerebrovascular system as well as the CSF flow are characterized by moderate Reynolds number and are thus mostly in a laminar regime. The complexity of the anatomical vessels, however, leads to fluctuations in the dynamics of these flows often resembling a transitional regime characterized by high-frequency fluctuations and dissipation of turbulent kinetic energy. A detailed assessment of such flows, down to the level of Kolmogorov scales remains one of our main research areas in applications like blood flow in intracranial aneurysms and CSF flow in Chiari malformation. Furthermore, we continue to set out computational benchmarks using simplistic geometries like a model stenosis and the FDA nozzle for steady, pulsatile as well as oscillatory flows.

    Instantaneous vorticity magnitude at 6 points in an oscillatory flow cycle across a bisecting plane in an axisymmetric stenosis for Re = 2100 (Jain, K. (2019). Transition to turbulence in an oscillatory flow through stenosis. Biomechanics and modeling in mechanobiology, 1-19).
  • High Performance Computing

    HPC is one of the important tools used by the group for detailed computations of flow in varied applications. The Multiscale and Multiphysics nature of biomedical applications requires the development of advanced numerical schemes and codes that can execute simulations on massively parallel architectures.

    To that end for our CFD simulations of flow and transport, we use the Lattice Boltzmann Method (LBM) due to its scalability on parallel architectures and easy representation of complex anatomical geometries. On one hand, some of our computations are conducted on 16 cores of a workstation for initial analysis, and on the other, we execute simulations utilizing between 8000 to 500’000 CPU cores of modern supercomputers like the SuperMUC, HazelHen, Juqueen and PizDaint. The ultimate goal is to arrive at reduced order mathematical descriptions of the Physiological systems.

    Codes and selected media:

    1.  The end-to-end massively parallel simulation toolchain, the APES framework available as open-source software
    2. Musubi LBM solver showcased by FZ Jülich as one of the highest scaling codes on Juqueen, running at almost 1 million concurrent threads.
    3. The project Simulating Transitional Hemodynamics in Intracranial Aneurysms at Extreme Scale listed by the Gauss Center for Supercomputing. 

dr. K. Jain (Kartik)
Assistant Professor