Lure of collaboration

Interdisciplinary work attracts top women computer scientists

At any gathering of computer scientists, Sarah Frisken is usually one of only two or three women in attendance. So when she was appointed a professor in Tufts’ computer science department in January, the difference was marked: Out of 15 people in the department, eight are women.

Carla Brodley © Kathleen Dooher

The number of women in computer science and the fact that they outnumber men—even if only slightly—is something unheard of at most universities. In fact, department members joke they’ll have to do some outreach to bring more men onto the faculty.

According to a 2002 study of diversity in science and engineering faculties at 50 top research universities, less than 11 percent of the computer science faculty are women.

“We look broadly to discover wonderful talent, so part of the story is why top women choose Tufts,’ said Diane Souvaine, professor and chair of computer science at Tufts. One reason for the department’s ability to attract exceptional faculty, she explained, is that “gwe tend to keep low boundaries between schools. And that is appealing┼h because it’s relatively easy to collaborate with other departments and researchers. Also, she said, “Tufts is small enough so that each person can make a difference, and that’s exciting.”

Every faculty member in computer science works in other disciplines. Frisken and Carla Brodley, also a recent hire, do research that has applications in medicine as does Lenore Cowen, who came to Tufts in 1998. Donna Slonim, an associate professor who started teaching in January, will hold a secondary appointment in the School of Medicine’s pathology department, pending final approval by the Board of Trustees.

But the department’s interests are broader than medicine.

Brodley works in the area of digital libraries, and Soha Hassoun, whose specialty is VLSI (Very Large-Scale Integration) microprocessing, was a chip designer at Digital Equipment Corp. before coming to Tufts in 1998. Souvaine investigates effective computation of new statistical analysis techniques.

Computer science and its applications are fundamental to research across all disciplines, Souvaine said. “Whether it’s music, the Fletcher School, biology, physics, classics, the veterinary school, it’s important to grow computer science and infrastructure for computer research,” she said. “For example, computers can be instrumental in organizing, querying and presenting data. Imagine wanting to answer questions such as ‘What is the water quality in the town a patient lives in? What pathology is being observed in pets in that area?’ Or imagine how to facilitate a drama student searching automatically through a video or a database of videos to find particular scenes?” Computers can address all these questions. In fact, Souvaine added, there has been discussion of establishing a Tufts University research network that would facilitate research collaborations and data-sharing across disciplines.

Electronic footprints
Here is a look at work being done by Carla Brodley, Sarah Friskin and Donna Slonim, the three most recent hires in computer science:

Sarah Frisken © Kathleen Dooher

Say you’ve logged on to your computer using your password and then leave your desk for a while. In your absence, someone else comes along, uses the computer and leaves before you get back. Could you tell someone else had been there?

The answer is yes, thanks to a program Professor Carla Brodley developed. Brodley, who joined the faculty last September, specializes in computer security research and in data mining, which she defines as “understanding data to make predictions or discover new things.”

Both areas are increasingly vital. Companies, governments and institutions are concerned about computer security, while data mining is part of a new field known as KDD, or knowledge discovery in databases.

Modern science is able to collect immense amounts of data—so large that it is impossible for humans to analyze all the information, Brodley said. As a result, scientists are looking for ways to automate data analysis.

Brodley worked on a project, for example, that uses data analysis to help diagnose lung diseases. She helped develop a method in which a physician, armed with a high-resolution CT scan of a lung, can send the image as a query to a database filled with other images of lungs identified by disease to compare the images and help make a diagnosis. Without this procedure, a doctor would have to manually go through image after image to make the comparison.

In the case of an unauthorized computer user, Brodley explained that everyone uses a computer in a unique way, leaving a kind of signature because every signal sent to a computer can be detected.

“There are patterns people use when they are typing,” she said. “People have different key stroke speeds. They use the mouse differently—some wiggle it around, and some don’t. People use particular commands.”

Computer-aided surgery
Sarah Frisken’s expertise is in computer graphics. Her research will allow surgeons to practice their techniques on a computer screen or on a three-dimensional model, making surgical outcomes more successful. Frisken spent more than 11 years as a researcher at Mitsubishi Electric Research Lab in Cambridge. An intern at the company who was a student at Tufts introduced her to Souvaine. “It seemed like a great opportunity to come here,” she said.

Frisken is teaching a course in computer graphics, which, she said, “has become quite a big field. I worked on surgical simulation—three-dimensional pictures of human anatomy. The vision is that this would supplement or even replace anatomy labs for teaching. We could build a patient-specific model and have a 3D scan, and surgeons could plan what instruments are needed and see what problems they might encounter.”

She also works in a field called haptics, the science of applying tactile sensations to human interactions with computers. Frisken has collaborated with researchers from several institutions on a project that would allow a surgeon to practice a procedure with a device that would feel the same as if he or she were using it on a knee, with the same force and resistance the knee would provide. “The device pushes back on you,” she said, “and it feels like you are actually doing this procedure.”

Cancer genetics
Before she came to Tufts, Donna Slonim was a scientist at Genetics Institute and at the Whitehead Institute Center for Genome Research. Her research involves developing a better understanding of what causes cancer. “Technology came along in the mid-nineties that shows which genes are being expressed in a sample of cells, which are being used to make proteins. This gave us an idea of what the cell is actually doing. We now have a molecular perspective on what the cell is doing globally that we never had before,” she said.

“With cancer, we don’t have a good understanding of what causes the changes at the molecular level, what causes cancer and what causes the destructiveness at that level. So now we potentially have the ability to look at what’s going on in these cells and compare it to what’s going on in cells that are not cancerous or are perhaps more responsive to a particular treatment, and make the distinction.”

In a given experiment, Slonim said, a researcher may monitor tens of thousands of genes. “Most of those thousands of genes may not be relevant to the problem, so most computational methods don’t deal with this well. My research deals in finding computational methods that can extract patterns from data sets of this type.”

She also does work in systems biology—understanding how biological systems work together and what goes on in a particular system—for example, how genes and proteins interact.

Slonim expects a secondary appointment in the pathology department at the medical school. “One of the things I’m really excited about is the opportunity to collaborate with scientists and medical doctors. I’m hoping to form an interdisciplinary group of students and researchers that spans both campuses.”

Marjorie Howard is a senior writer in Tufts’ Office of Publications. She can be reached at marjorie.howard@tufts.edu.