DiaGrid helps advance biomedical imaging for understanding and treating cancer and other diseases

  • April 28, 2015
  • Science Highlights

Purdue’s DiaGrid hub is helping advance a biomedical imaging technique capable of precisely capturing mechanical properties of bodily tissue, which could improve diagnosis and treatment of breast cancer and other diseases.

DiaGrid now houses a Web-based science gateway for the imaging technology, called NLACE. The hub lets researchers submit data easily for processing on national supercomputers and get results automatically. The research is a multi-institutional effort led by Paul Barbone at Boston University and Assad Oberai at Rensselaer Polytechnic Institute.

The Scientific Solutions Group in ITaP Research Computing developed the gateway in conjunction with the National Science Foundation’s Extreme Science and Engineering Discovery Environment (XSDE). Purdue is a partner in XSEDE, the largest collection of digital resources for research in the world.

The ITaP group can assist Purdue researchers in using XSEDE, as well as with computational application design and data set management; software development and consulting; and collaboration on grant proposals. For information contact Carol Song, senior research scientist, 49-67467, carolxsong@purdue.edu.

Numerous pathologies are linked to changing mechanical properties of tissue, from swollen glands and hardening of the arteries to cirrhosis of the liver and stiff lumps that can signal breast or other cancers.

Removing and testing tissue samples can determine mechanical property changes for purposes of diagnosing these diseases, but NLACE promises more precise measurement and a more complete picture without invasive sample taking.

Barbone likens coupling NLACE with ultrasound, magnetic resonance imaging (MRI) and other commonly used biomedical imaging technologies to adding infrared capability to a regular camera. Through the additional information NLACE generates, it can show things, like a budding tumor, that might be invisible to standard scanning methods alone.

“NLACE has been applied to soft tissues like organs and hard tissues like bone; it’s even measured mechanical property distributions within individual living cells,” Barbone says. “NLACE can infer not just linear, but also nonlinear properties; not just properties that are the same in any direction, but also directionally dependent properties that vary depending on the direction the material is pulled; not just how elastic the tissue is, but how fluid and viscous as well.”

Besides potential for integrating NLACE into clinical scanners, the research advances basic imaging science, along with understanding of how tissue mechanical properties play into development of diseases. In breast cancer, for instance, mechanical property changes may actually spur malignancy, notes Barbone, a professor of theoretical acoustics and applied mechanics.

“We used to think only that some diseases cause a change in the mechanical properties,” Barbone says. “Now we recognize that in some cases, changes in mechanical properties contribute to the disease. The mechanical properties of tissues surrounding a tumor, for example, can influence the biological behavior of tumor cells.”

Originally, the group envisioned NLACE as a downloadable Linux application. However, the collaboration involves more than a dozen biomedical labs. For them, installing and maintaining the software was going to be burdensome, and was likely to discourage others from trying it in the future. The plan is to make NLACE openly available to the research community.

The desire to lower the barrier to entry led to the idea of a Web gateway offering the software as a service.

Purdue computational scientists Chris Thompson and Lan Zhao were able to quickly develop the gateway by leveraging the DiaGrid hub and HUBzero. The Purdue-developed HUBzero platform lets researchers run research software from a Web browser without installing anything, while connecting seamlessly to supercomputers on the back end. DiaGrid, built on HUBzero, features a suite of computational tools for science, ranging from molecular dynamics to climate modeling, all available through easy-to-use graphical interfaces.

“I liked the fact that we didn’t have to reinvent the wheel,” Barbone says.

Through the NLACE gateway, users upload data that is then transferred for processing on Gordon, one of the XSEDE supercomputers, without any further intervention. The results come back in open source Visualization Toolkit (VTK) format. The gateway incorporates ParaView, an open source data analysis and visualization system.

Members of ITaP Research Computing’s Scientific Solutions Group, Thompson and Zhao also are part of the support team for XSEDE.

“We see more and more principal investigators like Dr. Barbone interested in making their codes, and the ability to run them, available to the public through science gateways,” says Nancy Wilkins-Diehr, co-director of XSEDE’s Extended Collaborative Support Service (ECSS).

The research to develop NLACE also requires the kind of high-performance computing XSEDE makes available. Essentially, the researchers are presented with a result and then have to computationally reverse engineer an optimal answer to why they got it, says Oberai, a professor and associate director of Rensselaer Polytechnic’s Scientific Computation Research Center. They run problems through hundreds, even thousands, of iterations to glean optimal solutions.

Originally posted: April 28, 2015 10:28am EDT