Mind over Mapping - Bioscience Technology Online

Bioscience Technology

100 Enterprise Drive
Rockaway, NJ, 07866

Mind over Mapping

by Mike May

This Atlas’ Long Road
The work behind NeuART started more than 20 years ago in the laboratory of Larry Swanson, Milo Don and Lucille Appleman Professor of Biological Sciences and professor of biological sciences, neurology, and psychology at the University of Southern California. Even Burns got involved with NeuART about a decade ago. It never quite reached maturity, though, and everyone moved to other projects. In 2001, Burns received funding to develop his NeuroScholar program that manages neuroscience information, and he wanted a neuroanatomical plug-in to use with it. So he resurrected NeuART. His programmer at the time, Wei Cheng Cheng, performed what Burns calls “the vast majority of the work that went into the design and implementation for the final system.”

Neuroscientists use published brain atlases to identify the location of things in the brain. “This is important within experiments since neurons in different places are involved in different brain functions,” says Gully Burns, research assistant professor at the Information Sciences Institute in the University of Southern California, “For example, visual cortex is very different from auditory cortex.” In many cases, scientists still draw these maps by hand, and a stack of drawings makes up a brain atlas. Comparing data within a stack proves difficult enough, and comparing data between stacks tends toward impossible. But Burns is working on a solution, one with a long history.

In general, Burns works on managing information. For instance, he runs USC’s NeuroScholar system, which helps neuroscientists manage information, including journal articles, research data, and even schematic diagrams. In addition, he works on NeuART, a project designed to solve the brain-atlas problem. As Burns explains, “Our system allows people to compare data within and between stacks of brain drawings so that they can compare maps from multiple experiments to look for patterns and similarities.”

In many ways, the real strength of NeuART comes from the ability to compare data from different experiments across atlases, which permits comparing data between different species. In describing NeuART, Burns says, “It’s a little bit like having a map of all the best sushi houses in a neighborhood and being able to overlay it on top of a map with all the best dance clubs. By joining the two data sets together you can plan a great evening out.” Turning to a more scientific analogy, he adds, “Similarly, if you had a map of lesion sites that cause effects in drinking behavior, say, and maps of the connections of the those areas, you’d be able to make testable inferences about the organization of circuits involved in drinking behavior.”

Simple coding for a complex challenge

In essence, NeuART organizes brain diagrams from pages in published books. Luckily, publishers provide a book’s plates as Adobe Illustrator files on an accompanying compact disk. NeuART converts those files to standard vector graphics, or SVG, format. The converted files can then be combined or compared.

NeuART’s key comes from input. Specifically, its development evolved from the input of expert neuroanatomists. “It is designed to meet their mapping needs,” says Burns. Doing that did not demand high-powered programming. “It’s a relatively simple Java software application,” says Burns. The program relies on open-source database and libraries to display data. He adds, “From a computer-science perspective, it’s nothing particularly fancy.”

Building computerized atlases based on material in books also generates another hurdle: copyright. “We’ve managed to overcome problems of copyright that block the development of informatics tools that can support atlas-based data,” says Burns. “This is essential since brain atlases are some of the most-cited documents in all of science.” To solve this problem, NeuART first checks the copyright on the CD, making sure that it is acceptable to use a book’s plates in this way, and it even checks to make sure that the user has a valid copy of Adobe Illustrator. With all of the checks confirmed, NeuART gathers up the plates, and can then arrange or compare them in essentially any way imaginable.

Miles of possibilities

NeuART can collect a wide range of neuroanatomical information, as shown here with data from the lateral septum and bed nucleus of the stria terminalis.

NeuART currently exists in a beta version. Still, Burns puts it out for testing. He says, “We’ve used it in meetings in Larry Swanson’s lab where someone will be discussing a particular terminal zone or region of the brainstem, and we can just pull up the maps and see exactly what the data is showing us.” He adds, “Other than that, we’ve just put it out there in the hope that other people find it useful.” Anyone interested can find out more at the NeuART website (http://www.neuroscholar.org/neuart2.html).

Eventually, Burns hope that NeuART will turn into what he calls a neuroanatomical knowledge acquisition tool. As an explanation, he says, “A lot of people doing neuroscience use atlases and do mapping work. What they don’t do is share their data.” He and his colleagues hope to develop a project in which people submit brain maps to a central database, he says, “where map-based research findings can be easily searched and synthesized,” all with the help of NeuART. Moreover, Burns hopes to give this database even more power. “We’re going to mine the literature using natural language processing,” he says, “so that we would be able to then link each fact from a given paper to the appropriate location in the brain within a graphical interface.”

In the end, NeuART could play a role in a broad collection of neuroscience information. In looking down the road to the central database collecting mapping and literature information Burns says, “This would be a little like Google maps for neuroscientific knowledge.”