Advanced Computing is Helping a Researcher Overcome a Barrier
Clockwise, from top left: Justin Baba, joint faculty associate professor at the University of Tennessee, Knoxville, and Research and Development staff member at Oak Ridge National Laboratory (ORNL); Dwayne John, staff member of the User Assistance group at the National Institute for Computational Sciences; and Vijay Koju, student in the Middle Tennessee State University Computational Science Ph.D. program and ORNL intern. [Image credits: Scott Gibson and Dwayne John]
Researcher Justin Baba of the University of Tennessee and Oak Ridge National Laboratory (ORNL) was encouraged by the results he had obtained from light-scattering experiments in turbid media because, in his view, they expounded on existing scientific literature on the subject.
An understanding of the behavior of light in turbid media could lead to the ability to non-invasively scan suspicious skin lesions to see if the hallmarks of melanoma lie beneath the surface, and the concept could be applied in other areas as well, such as being able to visualize objects hidden by fog, clouds, or murky water, for example.
While confident that what he had seen in his experiments could advance the science, peer reviewers of his research manuscript suggested he supplement his findings with computer simulations.
Baba—who has joint faculty appointments in the Civil and Environmental Engineering Department at UT Knoxville and at UT Graduate School of Medicine, Department of Surgery, and is an ORNL Research and Development staff member—says the reviewers’ suggestion represented a metaphorical wall, in that he had no experience using high-performance computing for modeling and simulation. What’s more, he adds, acquiring the necessary skills for the task would have taken him on his own, years.
The answer, he decided, was to collaborate with an expert in advanced computing he had met a few years back.
This is where Dwayne John of the User Assistance group at the National Institute for Computational Sciences (NICS) entered the picture. John, along with ORNL intern and Middle Tennessee State University student Vijay Koju, worked with Baba to develop a Monte Carlo simulation that enables the tracking of photons as they spread (propagate) or are absorbed in light-scattering media.
The NICS-managed Nautilus (now decommissioned) and Darter systems provided a means of probing assumptions regarding photon propagation. The machines facilitated timely propagation of sufficient photons while tracking the parameters of interest in the interactions, Baba says.
The team found that the quantum physics concept known as Berry phase is strongly associated with how the photons permeate the material. “Because of this, we see the potential for studying highly scattering samples a different way,” John says. “This method is polarimetric depth-resolved characterization. In essence, we are trying to relate the polarization of the photon with how deep the photon gets in the sample.”
Details of their discovery are contained in the conference paper “Monte Carlo based investigation of Berry phase for depth resolved characterization of biomedical scattering samples,” published in SPIE Proceedings on March 10, 2015. Baba notes that he and the team are also working on a manuscript for journal submission that will focus on explaining observations contained in various other published papers on light propagation in turbid media.
“A lot of studies have been published on the impact of detecting or tracking the polarization of photons,” Baba says. “But there isn't much understanding of what is actually occurring as these photons are scattering and propagating through the media. We have captured step by step what is taking place at each scattering event and the effects on the photons’ properties due to a scattering event. So we are able to elucidate what is already out there in the literature.”
The researchers embedded a metered 1950s United States Air Force imaging target in scattering media and, through the use of Monte Carlo simulations, generated an image of the target.
Animation composed of reconstructed images of a U.S. Air Force metered target. The images become sharper as the depth of the placement of the target decreases in solution. The solution, made up of micron-size polystyrene spheres suspended in water, represents a “natural scattering medium,” such as milk or biological tissues. The depths in the solution are 1.5cm, 1.0cm, 0.75cm, 0.5cm, 0.4cm, and 0.2cm.
Baba explains that the most robust aspect of the research so far is that it is couched in sound physics and that it concurs with the commonly held belief that examination of the circular rather than the linear polarization component of the detected photons provides more information in turbid media at a greater depth.
In terms of what gives him pause, Baba says it is the sheer amount of data produced. Whereas the standard maximum number of propagated photons reported on in existing literature is in the millions, Baba’s research is propagating trillions of photons, resulting in more data than can be stored and retrieved. Consequently, much of the data has been discarded during the studies.
“In the future, we’ll rerun the simulations with different parameters,” he notes. “That’s unfortunate [to have to give up data], but the computational aspect is the greatest challenge. It takes a significant amount of computational power and resources to run the simulations, compile the results, and analyze those results well.”
The right expertise, Baba says, was crucial to tapping into the power of supercomputing to advance the science: “Vijay has been handling the coding and running the simulation studies, and Dwayne has been handling the sequencing of our runs and helping to figure out how to convert the code that contains the physics into the format that will run and function on the high-performance computing resources.”
Scott Gibson, science writer, NICS, JICS
Article posting date: 21 October 2015
About JICS and NICS: The Joint Institute for Computational Sciences (JICS) was established by the University of Tennessee and Oak Ridge National Laboratory (ORNL) to advance scientific discovery and leading-edge engineering, and to further knowledge of computational modeling and simulation. JICS realizes its vision by taking full advantage of petascale-and-beyond computers housed at ORNL and by educating a new generation of scientists and engineers to be well versed in the application of computational modeling and simulation for solving the most challenging scientific and engineering problems. JICS operates the National Institute for Computational Sciences, NICS, one of the nation's leading advanced computing centers. NICS is co-located on the UT Knoxville campus and ORNL, home of the world's most powerful computing complex. The center's mission is to expand the boundaries of human understanding while ensuring the United States' continued leadership in science, technology, engineering, and mathematics. NICS is a major partner in the National Science Foundation's eXtreme Science and Engineering Discovery Environment (XSEDE).