Engineers, says Karen Panetta, enjoy seeing their accomplishments used in the real world. "Theorists might enjoy formulation and mathematical rigor. But as an engineer, you want to see it in practice, improving the quality of peoples' lives."
Using the technology being developed by Karen Panetta, airport screeners could not only see the outline of objects, but identify what they are made of. Photo: Tia Chapman
Panetta, an associate professor of electrical and computer engineering, may soon have that satisfaction. She has developed a new image-processing technology that is faster, better and less expensive than what currently exists.
It would allow airport screeners, for example, to see not only the outline of an object in a suitcase, but discern what it's made of. And, where a doctor once saw a tumor on a conventional mammogram, she might now be able to see where cancer from that tumor has spread.
"It's an imaging technology that can enhance and bring to the human eye things that aren't visible in just the output from a camera," says Panetta. "It has the ability to revolutionize image-processing techniques for security applications as well as for biomedical imaging." One of the goals, she says, is to develop an automated detection system because it would be more accurate and discerning than the human eye.
To make it work, she's employed something not at all revolutionary-Boolean algebra-in a new way. "The key is reducing the complexity of computation involved in processing digital images," she says.
Simply put, when you work with huge image files, it takes a long time to load them-that's computation time. "What we're doing works on very large images with very inexpensive equipment, so you can do it for pennies as opposed to using much more expensive chips," Panetta says.
The fields of digital electronics and image processing have much in common, but have mostly developed separately. Panetta used her knowledge of both to devise the new ways of using Boolean algebra-that's the "and," "or" and "not" you see in some advanced Web search engines-to come up with faster ways of compressing data.
"What's happened is that the fields of image processing and digital electronics never talk and haven't utilized the techniques on each side," says Panetta. "The things we're doing are very straightforward, but no one made the connection because people don't think beyond their own disciplines." Panetta says her work in the computer industry-she worked for Digital Equipment Corp.-and her current research in image processing helped her apply knowledge from each area.
Allied Minds Inc., a Quincy, Mass., investment firm that provides seed money to projects being developed by universities and laboratories, learned about Panetta's idea from Tufts' Office of Technology Transfer and decided to invest in her work. The company BA Logix, the first start-up generated from the School of Engineering, was founded to develop and market the concept.
Company revenues will be shared by Tufts, Panetta, Allied Minds and BA Logix. Seed capital provided by Allied Minds includes stipends for graduate students working on the project.
Martin Son, associate director of the Office for Technology Licensing and Industrial Collaboration, said Tufts has applied for a patent on the technology, which BA Logix will eventually license. "Once in a while a technology is broad enough that it makes sense to start a company around it, and that was the case here," Son says.
The technology will also be useful in data compression or "any application where a large amount of data has to be transmitted," says Vincent Chun, a director at Allied Minds. Data compression refers to methods of sending information that otherwise would take up a lot of bandwidth. "Karen's method is fast, and it requires less computation," says Chun.
Panetta says she is also pleased that the project is involving her students. "They will get to see their work realized in real products, and they are serving a real client," she says. "There's no walking away with a B+ on this one. It has to work. It has to be an A+."
Marjorie Howard may be reached at firstname.lastname@example.org.