A macro-scale model of the knee (left) was created to study compressive loading of the joint. At the micro scale, the single-cell model (top, right) has been used in previous studies, while Erdemir’s 11-cell model better represents the effects of loading on the individual cells. Images: Erdemir/Cleveland Clinic
Cleveland Clinic research team is developing virtual models of human
knee joints to better understand how tissues and their individual cells
react to heavy loads—virtual models that someday can be used to
understand damage mechanisms caused by the aging process or debilitating
diseases, such as osteoarthritis.
by Ahmet Erdemir, Ph.D., the team is leveraging the powerful computing
systems of the Ohio Supercomputer Center to develop state-of-the-art
computational representations of the human body to understand how
movement patterns and loads on the joints deform the surrounding tissues
and cells. Erdemir is the director of the Computational Biomodeling
Core (CoBi) and a faculty member in the Department of Biomedical Engineering at the Lerner Research Institute (LRI) in Cleveland, Ohio.
aging process and debilitating diseases affect many aspects of the
mechanical function of the human body: from the way we move to how our
muscles, joints, tissues, and cells accommodate the loading exerted on
the body during daily activities,” Erdemir explained. “Computational
modeling techniques provide an avenue to obtain additional insights
about mechanics at various spatial scales.”
macro-scale studies have looked at how the various components of a knee
joint—cartilage, menisci, ligaments and bone—respond to weight and
other external loads. However, Erdemir and colleague Scott C. Sibole
wanted to better understand how those large mechanical forces correspond
to the related deformation of individual cartilage cells—or
chondrocytes—within the knee. Previous micro-scale studies of cartilage
have not commonly been based on data from body-level scales, in
particular, by the musculoskeletal mechanics of the knee joint.
addition, calculated deformations typically have been for a single cell
at the center of a 100-cubic-micrometer block of simulated tissue;
Erdemir used an anatomically based representation that calculated
deformations for 11 cells distributed within the same volume.
Erdemir’s finite element model of the knee joint with representation of the cartilage, menisci and the associated bone structures. An enlarged model region (right) illustrates the mesh resolution of the simulation.
both micro-scale approaches, the cartilage cells experienced amplified
deformations compared to those at the macro-scale, predicted by
simulating the compression of tissues in the knee joint under the weight
of the body,” Erdemir found. “In the 11-cell case, all cells
experienced less deformation than the single cell case, and also
exhibited a larger variance in deformation compared to other cells
residing in the same block.”
method proved to be highly scalable because of micro-scale model
independence that allowed exploitation of distributed memory computing
architecture. As a result, Sibole, a research engineer at LRI, was able
to leverage the computational muscle of OSC’s IBM 1350 Glenn Cluster. At
the time, the 9,500 nodes of the Glenn Cluster provided 75 teraflops of
computing power, tech-speak for 75 trillion calculations per second.
Recently, the Glenn Cluster was partially decommissioned when engineers
deployed the center’s more powerful HP-Intel Xeon Oakley Cluster.
of OSC’s two most recent flagship computing systems were specifically
designed to support biomedical applications, such as those employed by
Dr. Erdemir and Mr. Sibole,” said Ashok Krishnamurthy, OSC interim
co-executive director. “Researchers working at Ohio’s various respected
medical centers are conducting an ever-increasing load of computational
studies and analyses, and they now represent a significant share of our
Chondrocyte Deformations as a Function of Tibiofemoral Joint Loading Predicted by a Generalized High-Throughput Pipeline of Multi-Scale Simulations
Source: Ohio Supercomputer Center