The Unseen Challenge

By Joanne Silver

David Cyganski with the "mantenna" directional radio homing device, an offshoot of the personnel locator research.

They probe the ocean's depths and the innermost recesses of the mind. They wield wands and magnetic coils and designs pulsing with vibrant colors. Working in fields ranging from neuroimaging to machine vision, WPI engineers and scientists are finding ways to uncover information hidden from plain sight and understanding. David Cyganski, Susan Jarvis, Reinhold Ludwig, and Matthew Ward are engaged in technological versions of hide-and-seek, chasing the elusive knowledge that could ultimately make a difference in the lives of humans and other beings.

When it comes to seeing the unseen, television and movies tend to romanticize and distort. There's no chance, explains David Cyganski, professor of electrical and computer engineering, that Jack Bauer of television's "24" fame could use his cell phone to call up indoor schematics identifying enemy locations. Cyganski should know. He is part of a team that has been laboring for years to develop a tool for detecting people inside a building.

Accurately locating and tracking people in indoor environments is a daunting technical task. GPS, capable of precision location outdoors, is notoriously ineffective inside buildings, where signals from the GPS satellites bounce off walls, losing some of their accuracy.

Cyganski and graduate student Vincent Amendolare '06 discuss prototype technology for the locator system. The screen display shows the results of a system test; the dark red area represents the likely location of the person being tracked.

Cyganski knows that for the personnel locator to be effective, it must be able to pinpoint the location of a first responder and trace the responder's path, in three dimensions, to within about a foot. Such accuracy is necessary to determine, for example, on which side of a wall a firefighter is standing. A small number of university, corporate, and government labs are tackling this problem utilizing different approaches. WPI's strategy appears to be among the most promising.

The system combines principles from OFDM (orthogonal frequency division multiplexing)—a form of high-speed data transmission—and techniques that can be employed to extract great detail from radar signals. Responders will wear badges that will continuously transmit OFDM signals. Receivers on fire trucks positioned around the building will pick them up and use custom-designed algorithms to sort out the straight-line and reflected signals to determine the exact location of the transmitters.

Over time, the accuracy and range of the technology have steadily increased from the initial prototype, which had a range of 50 feet and an accuracy of about 5 feet. Today, the large, well-funded ($3 million, to date, from the U.S. Department of Justice), multi-specialty team of faculty members and students is getting closer to that one-foot goal, along with two other key objectives: a 2,000-foot range and the capacity to track up to 100 people simultaneously.

Surrounded by books on fiber optics and data networks, red WPI binders, photographs of his son, a poster of Batman, and an array of coffee mugs, Cyganski speaks movingly about his current endeavor. The project began in response to the 1999 fire at the Worcester Cold Storage warehouse, according to the professor whose longtime ties to Worcester include three WPI degrees. "We didn't think it was right," he explains with tears in his eyes, recalling the deaths of six firefighters. All had succumbed within feet of exits they couldn't locate in the dense smoke.

Reinhold Ludwig with the MRI breast coil. The coils fit into a mechanical holder on which the patient is placed prior to scanning.

A Better Way to Detect Breast Cancer

For Reinhold Ludwig, too, feelings have intensified his scientific commitment. In his search for coils that can be used to produce high-resolution MRI images for breast cancer detection, he thinks of his grandmother, who died of the disease. He says, "If you can see the early onset of cancer, you can be more successful. You get personally involved. It's not enhancing a missile control system. It's enhancing a person's life."

Years of developing and refining magnetic coils for MRI scanners have given this electrical and computer engineering professor ample opportunity to observe the mysteries of life—from the signature of fear in a rodent brain to the first speck of cancer in a human breast. In each case, he employs his engineering perspective to improve the underlying magnetic resonance hardware in order to generate better images.

In MRI, a combination of a powerful magnetic field, so-called gradient fields, and radio frequency (RF) fields, all produced by magnetic coils, are exploited to obtain voltage responses from the stimulated hydrogen nuclei in the human body; these signal responses are processed to form highly detailed anatomical images. Ludwig's RF coils are designed to enhance the imaging capability of existing MRI instruments by targeting specific regions in a human patient or in a lab animal.

These images, showing the breasts and surrounding tissue of a volunteer, were generated in a clinical GE Signa 1.5 Tesla MR scanner at Harvard's McLean Hospital.
Click here for a larger version.

His recently patented breast coil is anatomically shaped to focus the imaging capability on just the breast and sur-rounding tissues. When his two coils are interfaced into standard MRI scanners, they can simultaneously create high-resolution images of both breasts, along with the underarm lymph nodes, which are not visible in typical mammograms but need to be checked for the possible spread of cancer.

There is growing evidence that MRI is more accurate than mammography, but it is currently too expensive for routine cancer screening. (MRI breast exams are usually ordered only after a possible cancer has been detected with mammography.) Ludwig says his new RF coil concept, which focuses the MRI scan on a limited area of the body, could speed up the MRI exam, making it less costly. In addition, the images obtained with this coil are far superior in resolution and coverage when compared with existing RF coils in clinical MRI instruments.

Making Complex Concepts Easier to Visualize

Clockwise from top left, Matt Ward (at left) with some of the products of the visualization tools he develops to extract meaning from massive databases. The images present data from a biodiversity study (low-elevation clusters in orange); data on cars (four-cylinder cars in red; similarly shaped glyphs represent cars with similar attributes); crime data from Detroit (the correlation between high homicide rates, low cleared homicides, and high numbers of government workers is highlighted); remotely sensed data from Western Australia; and more car data (high-MPG cars in red).
Click here for a larger version.

In a nearby computer science lab, the human mind poses another set of hurdles for Matthew Ward. His XmdvTool is well crafted for the interactive analysis of large multivariate data sets. But he knows it would be of little value if it did not present information in a manner that grabbed people's attention. "Visualization draws a lot from art," the computer science professor says. "You want to make things aesthetically pleasing so that people won't mind looking at them." The tool is flexible enough to handle data related to such diverse areas as the stock market, space science, and warfare simulation in a way that makes sense to a variety of individuals. "It depends on who the user is," Ward says. "A trained map reader in the army will see features better and remember more." To allow those with less expertise to utilize his visualization tools, Ward concentrates on abstracting huge quantities of data into comprehensible, manipulable graphics.

Long before developing interactive visualization tools to expose patterns in vast databases, Ward enjoyed solving mathematical problems his father gave him. Plus, he says, "I tend to think visually. I look at problems and try to come up with visual analogies." Even now, when the situations he addresses might involve millions of pieces of information, Ward aims for solutions that are simple, elegant, and useful. In contexts as dissimilar as homeland security and genetic sequencing, the eye-popping punch of his displays makes mountains of unwieldy data a pleasure to behold.

In theory, there is almost no limit to the possible applications of Ward's visualization tools. He speaks of "keeping it simple and making it portable. It should be immediately off-the-shelf usable." Sitting by his computer, he displays two examples: a data set of information about cars; and factors potentially correlating with crime in Detroit. Origami shapes in periwinkle, red, and green ebb and flow according to the questions Ward poses. Before long, hundreds of thousands of random bits of data release their secrets, in the form of glyphs and networks of lines. Some are puzzling, such as why the number of homicides should rise with the growth of the government work force. But just to have these questions to ask is a sign of the growing success of the XmdvTool.

By applying the concepts of information visualization to such varied topics as astronomy and fire simulation, Ward and his students have made important inroads into an increasingly vital method of communicating information. His efforts at supervising graduate students helped WPI win a major award from the U.S. Department of Education. The XmdvTool project has been funded by the National Science Foundation since 1998, and has received funds from the National Security Agency.

Right, technology developed by Susan Jarvis is able to track the movements of marine mammals in real time by overlaying data from underwater sound sensors on map data in Google Earth. The colored icons represent different marine mammal species.

Probing the Water for Meaningful Noise

In a game of hide and seek, one only has to locate the unseen person. Susan Jarvis hasn't stopped at the initial discovery. Instead, she fine-tunes her methods and technologies in pursuit of the most precise—and thus valuable—solution.

This adjunct instructor in electrical and computer engineering credits a TV favorite from her childhood, "The Undersea World of Jacques Cousteau," for inspiring the focus of her adult career. Having worked in such areas as high-data-rate underwater acoustic telemetry, she now concentrates on developing methods to study marine mammals by using their vocalizations. Armed with information about animal movements and behavior deciphered from the clicks and whistles of whales and dolphins, she explores strategies to prevent the strandings that may be linked to the use of sonar by Navy vessels. She might not sound like Cousteau when she describes her mission—"We want to be able to do acoustic assessment of marine mammal activity in the presence and absence of anthropogenic noise"—yet her dedication to the water's inhabitants bears a striking resemblance to that of her television hero.

Having spent well over a decade investigating undersea acoustics, Jarvis still has many more questions than answers about the effects of ocean sounds on mammals. "The ocean in the last 50 or 60 years has become a very noisy place," she says. "Is that a big deal? Or has it happened gradually enough that it's not a problem?" The scientific community has too little quantitative information about the effects of man-made sound on marine mammals for researchers to understand fully what the impacts may be, she says. Jarvis describes an event that took place in the Bahamas in March 2000: "Ships had their sonar on. They were putting a loud sound in the water. As the ships went through, there were a number of strandings of beaked whales. Did it scare them? Did it knock them out of their normal behaviors? Was there cause and effect? It seems like a smoking gun, but sometimes there are beachings and sometimes there are not. We can't prevent the event unless we know what we are doing to perturb the creatures."

Jarvis's extensive efforts in developing new methods and tools to find and monitor marine mammals in the open ocean became a necessary step on the way to probing the bigger mystery of the strandings. In 2002, Jarvis tracked sperm whales by calculating how their clicking sounds would reach different undersea sensors aligned in a grid in one of the Navy's testing facilities. "Nobody was more surprised than we that it actually worked," she says.

Then came tests for other species of mammals. With newfound data about the types of sounds made, Jarvis says, "I've been doing a lot of algorithmic design" to track and classify an assortment of whales and dolphins. Now, she and her colleagues are hoping to get density estimates for different species, amounting to a kind of watery census. They also want to retrieve spatial and temporal information about fluctuations of populations over time—all with the goal of protecting the animals from harmful noise.

Despite a panoply of technological paraphernalia, these four professors harbor a very basic desire: to make life better for those inhabiting this planet. Two centuries ago, far from the gadgetry of modern laboratories, Henry David Thoreau wrote, "We are as much as we see." With the assistance and research of WPI engineers and scientists, we can see even more—and perhaps become more, as a result.