By Marlene Cimons, National Science Foundation
The computers in your doctor’s office these days might not be all that different from the one your child uses to play his video games: general purpose computers designed for a wide variety of applications. This is all well and good, unless your doctor needs to analyze your blood flow in a hurry.
Researchers at UCLA are designing new targeted types of computer software and hardware --a field known as domain-specific computing--to develop a health care computing prototype that could enable physicians to use computers in speedier, cost-efficient and much more focused ways.
With this new technology they could, for example, see inside the brain, quickly find a blood clot or analyze the results of an MRI or CT scan more rapidly than is currently possible.
“Today, it takes hours, hopefully, with our prototype it could take minutes,” said Jason Cong, professor of computer science and director of the new UCLA Center for Domain-Specific Computing (CDSC), which will oversee the research. “These computers will take an existing image and do the analysis.” It won’t replace the need for X-ray or a radiologist, “but will greatly facilitate the process,” providing a faster and more detailed interpretation, he added.
The work is supported by a $10 million grant over five years from the National Science Foundation as part of the American Recovery and Reinvestment Act of 2009.
“The scope of our project is to demonstrate the feasibility and benefits, and then, hopefully, the computer industry will build them,” Cong said.
Currently, most computers are general purpose systems, built to respond to increasing demands in various fields. In the present scenario, tens of thousands of computer servers are connected in warehouse-scale data centers. But they face serious challenges in terms of performance, energy, space and cost.
Cong and others believe that domain-specific computing, which uses “customizable” hardware and computer languages tailored to a particular application, will use less energy and produce faster results. The researchers also believe it will cost less, and increase productivity.
They are designing the prototype to serve health care needs, specifically medical imaging and hemodynamic modeling, which refers how the blood flows through the body’s blood vessels, but predict that, if successful, it could have much broader applications.
“It will be able to adapt to numerous applications, for example, the atmosphere,” Cong said. “It could be used to interpret global warming modeling. This will open up a new way of doing computing, and create a lot of new opportunities. It will likely create new jobs in the computer industry; if you have a new way of designing computers, you’ll have new possibilities for designing new hardware and software.”
With regard to medical imaging and hemodynamic modeling, “we’ll be able to see inside the brain and facilitate real-time surgery, for example,” Cong said. “Also, doctors will be able to do preventative procedures much faster with automatic analysis and diagnosis of MRIs and CT scan images. With hemodynamic modeling, the computer will be able to analyze the pressure, and tell you how thick the vessels are, or see whether you have a clot. We think there can be considerable cost savings in health care as a result.”
The broader impact will become part of “the new digital revolution enabled by customized computing,” he added. “We will demonstrate the feasibility and advantages of the proposed research in the domain of health care, given its significant impact on the national economy and quality-of-life issues.”
Furthermore, it will be necessary to train a new generation of students who are prepared for customized computing and who can effectively apply such techniques to many areas of society, furthering the digital revolution, Cong said.
With that in mind, the CDSC plans to integrate its research with education, exposing graduate, undergraduate and high school students to the new concepts and research from this project through several new courses jointly developed and shared by researchers.