Karen Panetta became a full professor in the Department of Electrical and Computer Engineering in 2009. Her current research focuses on improving health and safety through enhanced detection of such divergent threats as cancer, explosives, and aircraft failures. Enhanced image recognition systems will improve detection of cancer in mammograms and of explosives in luggage. Enhanced fault simulations will improve detection of potentially deadly electronic failures in aircraft. Speaking of her work, Panetta reflected, “It has to have some purpose in life. It’s got to be a benefit to humanity.”
Panetta earned her BS in computer engineering from Boston University and her MS and PhD in electrical engineering from Northeastern University. Prior to becoming a Tufts faculty member in 1996, she worked at Digital Equipment Corporation (Hudson, MA) and General Scanning (Bedford, MA). She continues to consult for MACOM/Tyco Electronics (Lowell, MA). Panetta is founder of Nerd Girls, an engineering outreach program, and is co-founder and chief research scientist of BA Logix, Inc., a startup company based on intellectual property developed in her research.
Panetta’s research has often been shaped by catastrophic events that highlighted a need for new products. Her simulation advances were spurred by the space shuttle disasters and X-ray machine accidents. “They could not determine what caused the malfunction,” says Panetta. “They had all these individual variables, and [didn’t know how to] simulate every possible condition.” So Panetta came up with multiple domain simulations that looked at hundreds of thousands of interactions among conditions that come together and could cause failure, in order to predict and prevent future disasters.
The terrorist attacks on September 11th stimulated a research shift into homeland security. “That was really my catalyst for going into image processing,” says Panetta. “I said, ‘This should not have happened.’” She started working on enhancement techniques that would allow the human eye to see the detailed information that is currently not visible in security camera images. She is also working on baggage screening for detection of explosives.
Her multimedia research can be compartmentalized into enhancement, detection, and materials identification. “In the enhancement portion, that means taking images from cameras and being able to see better what’s going on,” explains Panetta, who gives the example of a picture of her office with the window behind her. “You can’t necessarily see what’s outside the window, but with enhancement techniques, you would be able to visualize and actually see what’s outside, beyond the window.” Her group works on techniques that enhance data so the eye can see previously unseeable details. “So our entire research is based on modeling the human visual system,” says Panetta. “When you take a picture, what your eye can actually see is based on the limitations of the vision system. The information is there. How do we best extract it so human beings can see it?”
Panetta’s research is also focused on improving cancer detection. “To see things like mammograms, and to have someone say ‘You’re fine,’ but then to go back several years later and have them say, ‘Oh, there was something there. Somebody just couldn’t detect it.’” Panetta thinks that is unacceptable, so she and her research group are working with mammogram images to figure out new computer algorithms that will improve detection.
Other projects include improving the security of images transmitted over the internet and developing a database that collects and analyzes information on physically and mentally disabled people in India. Institutions and organizations need data about people affected by diseases in order to develop programs to address their problems. Data include an individual’s age, sex, address, and health history. “Sam [one of her students] came up with the secure database,” says Panetta. “All the schools and social systems put the information into their systems, and it’s all done for free, all done with open source software. It was launched last year—four million users and it hasn’t gone down once—that’s when you learn good design practices. The next step is to automate this into handheld devices, which Sam is working on now.”
For more information, please go to http://www.ece.tufts.edu/~karen.