The new age of bioimaging

Digital microscopy and powerful software are turbocharging systems biology

“Anyone can take pretty pictures,” says research scientist James Evans, who oversees the imaging arm of MIT’s Computational and Systems Biology initiative. “The challenge lies in extracting information from them and telling a story.”

Take the case of Whitehead postdoctoral fellow Robert Wheeler, who generated more than 100 gigabytes of data this spring by taking pictures of yeast cells. A computer program scanned all the images and identified proteins that mask fungal pathogens from our immune system, shedding light on the action of certain antifungal drugs.

Welcome to the era of digital microscopy, defined by quantitative analyses rather than qualitative observations.

Though the first multi-lens microscopes were built about 400 years ago, and the devices have been a mainstay of biological research nearly as long, scientists couldn’t harness their full power until digital photography made it possible to leverage computational tools for image analysis. Researchers now use automated microscopy systems to take thousands of pictures of cells, and software to mine the pictures for patterns.

Paul Matsudaira, Whitehead Member and director of the Whitehead-MIT BioImaging Center, believes this new form of bioimaging will play a powerful role in the emerging field of systems biology.

“Bioimaging will help scientists probe the complex relationships between genes, proteins, cellular components and physiological systems,” he says. “Systems biologists need tools to collect and make sense of mountains of data.

But it’s still early days for digital bioimaging. Few labs use quantitative characterizations of pictures to advance their research. Microscopy pioneers blame a variety of factors, including the high cost of cutting-edge microscopes and data storage. (The Whitehead-MIT center can store more than 40 terabytes—about 10 million pictures taken with a consumer digital camera.)

Two of the biggest challenges, however, are the need for better image-analysis software, and for a greater awareness among life scientists about how these new tools can benefit research.

Software with a sharper focus

Mining a gigabyte of data is no picnic. Imagine sifting through thousands of photos in search of cells that are dividing, or cells containing a specific fluorescently labeled protein. The image-analysis software that’s needed doesn’t materialize on its own. Scientists must write or tweak computer programs virtually every time they design experiments involving advanced microscopy.

“Bioimaging software is more powerful and easier to use than it was a few years ago, but it’s not plug and play,” says Matsudaira.

Biologists didn’t invent image-analysis software. The federal government first used computer programs to locate missiles, tanks and ships in photos taken by satellites during the late 1950s.

During the last decade, researchers began applying the descendants of these early programs to biology, training computers to recognize cellular components. Bioimaging software improved as scientists modified beta versions to meet their needs.

A growing group of researchers, including postdoctoral fellow Anne Carpenter of Whitehead Member David Sabatini’s lab, contribute to this iterative process.

Dissatisfied with existing software, Carpenter initiated collaborations with scientists in several labs at Whitehead and MIT and created a program that allows them to test the effects of many genes on cell size and appearance. “Coming from the biology side, I knew what the software needed to do, and I knew enough programming so I could get started on the project,” she says. “I turned to computer science graduate students for help when I got stuck.”

The software is freely available and can analyze a wide variety of biological images, but investigators must adapt it for specific cell types and conditions.

When Matsudaira and former Whitehead Member Peter S. Kim (now president of Merck Research Laboratories) first conceived of the Whitehead-MIT BioImaging Center in the late 1990s, they recognized the importance of customized software. The first hires were Evans (a molecular cell biologist who specializes in imaging), computer systems engineers and other scientists with image-analysis and computation expertise.

“When scientists approach me with projects, I can’t pull out a cookbook and tell them what to do,” says Evans. “We work together to identify the parameters and stitch together the appropriate tools. The technology develops through these collaborations.”

Matsudaira believes the pace of development will increase as biology educational programs place greater emphasis on computation. Graduates will feel comfortable working with complex equations and enhancing software on their own.

Spreading the word

The number of labs involved in digital bioimaging will also grow as scientists realize how it applies to their work. Professors associated with the Whitehead-MIT center expedite the process by designing demos to educate their colleagues. Each demo explores a real biological process, letting researchers answer interesting questions and publish their methods and findings in peer-reviewed journals.

Douglas Lauffenburger’s group, for example, used imaging to carefully map the movement of breast cells after over-expressing a receptor associated with cancer. The study, which could have implications for drug development, appeared in Biophysical Journal this August.

“Basic cell biology and pharmaceutical industry efforts currently don’t place much emphasis on detailed, quantitative characterizations of cell motility,” says Lauffenburger, who is director of MIT’s Biological Engineering Division. “It’s up to bioengineering labs like mine to demonstrate how important this is.”

That importance will only rise as the scale of experiments increases in many labs. Automated microscopy and image analysis should fare well as “big science” progresses.

The vignettes accompanying this story highlight the work of researchers at Whitehead and MIT breaking new ground in bioimaging.

*******************

Seeing the true image

Digital imaging holds great promise for research advances—and for abuse. Nicki Watson, who manages the W. M. Keck Biological Imaging Facility at Whitehead, trains scientists to avoid common pitfalls.

“It’s very easy to take a digital picture of what you want to see, but you need to discipline yourself to take a picture of what’s actually there,” she says.

Scientists sometimes run into trouble if they make mistakes while preparing specimens, alter images by hand and/or interpret them incorrectly. Photo-editing software, for example, allows them to easily eliminate noise and enhance elements of interest. But what exactly is “noise”? In some cases, the background of a photo contains relevant information.

“Scientists have an obligation to report how they’ve manipulated data,” says Whitehead-MIT BioImaging Center director Paul Matsudaira. He also urges scientists to avoid being led astray by image-analysis software. They should make sure such programs function as intended and keep in mind they could be missing some interesting patterns.

CONTINUED  1  2  3  4  5  6  7  Next >