Maryann E. Martone and colleagues at the University of California, San Diego, reported on their development of the Cell Centered Database at the Microscopy and Microanalysis meeting in Fort Lauderdale, Fla., in August. The database enables scientists to compare microscopic images of biological tissues using light and electron microscopy. Although the database contains image data from diverse organisms and tissues, many of the datasets are from the nervous system.

Martone said that the database was created to enable studies of neural networks, just as genomic data have enabled studies of genomes. She said that brain images contain information beyond the original purpose for which they were collected, and most of the information is left unused. The database allows neuroscientists to mine the imaging data for information that often is overlooked.

Martone’s group created and maintains the database, but the image data comes from other groups. The database originally was intended to be entirely public because it was developed as part of a National Institutes of Health initiative — the Biomedical Informatics Research Network (BIRN). However, most scientists initially want to keep their information private so that they can publish papers on it. Therefore, Martone’s group has made it possible for database users to manage private data, in addition to accessing the data that is publicly available through the network Web site.

The database not only includes images but also comprises reconstructions and segmentations derived from these images. Additionally, it provides the experimental context in which each image was taken, such as the methods used to fix the sample to the slide, because experimental procedures can affect observations made based on the data.

The brain images are spatially aligned according to coordinates from the George Paxinos brain atlas, a resource universally used in the neuroscience community. Through software under development as part of BIRN, the light and electron microscopy brain data from the database are being integrated with data from MRI and from histological studies. Because it is difficult to compare imaging modalities as diverse as MRI, Nissl stains and transmission electron microscope images, the database relies on user expertise to compare features among images. Martone said that it has been difficult for database experts and biologists to communicate using language both can understand, and that they constantly strive to make the process as easy as possible.

The researchers also have created the OntoQuest Cellular Knowledge Base, a tool that permits users to query the database, which will be made public by the end of this year. The search tool understands the data in terms of ontologies such as the following: a brain comprises neurons, which contain synapses. Users can employ the tool to find various neuronal features, such as all GABAergic neurons.

The tool enables category searches as does eBay or Amazon, a user-friendly feature, Martone said. The researchers are striving constantly to make the database user-friendly. They rely on user reports to let them know when a feature of the database is challenging to use.

Martone noted that most database users have been computational neuroscientists who apply algorithms to the data. For example, the database has been used to develop a computational model of the ciliary ganglion and to do fractal analysis of astrocytes.

So far, the researchers have worked mostly with confocal, multiphoton and transmission electron microscopy images. “We have plans to expand the schema to add new techniques,” Martone said.