19^th International CODATA Conference
Category: Data Visualization

Spatial integration of distributed heterogeneous brain images and segmentations

Ilya Zaslavsky (zaslavsk@sdsc.edu)
San Diego Supercomputer Center, University of California San Diego, USA

Haiyun He (hhe@sdsc.edu), Maryann E. Martone (mmartone@ucsd.edu), Amarnath Gupta (gupta@sdsc.edu)
National Center for Microscopy and Imaging Research, UCSD, USA

Integrated analysis and visualization of multiple representations of brain morphology and function is an important component of modern neuroscience. Online atlases of the brain are typically developed for 2D or 3D visualization of a single pre-built collection of images. We describe a novel atlas environment, which assists neuroscientists in discovering, retrieving, querying and overlaying distributed heterogeneous brain images and segmentations. The Smart Atlas being developed within the NIH-funded BIRN (Biomedical Informatics Research Network) project, comprises a collection of brain spatial data services registered on the grid, middleware components that support source registration, discovery and query, and a Web client used to formulate various queries across distributed resources and visualize composite query results.

The main challenges of spatial federation of brain data derive from the need to convert disparate brain images and markup into spatial databases that are accessible over the Internet via a standard set of location-based and attribute-based queries, and to develop a system for defining and managing integrated views over distributed heterogeneous brain data services.

We have registered several mouse brain image collections (from BIRN-affiliated labs at Harvard, Duke, UCLA, CalTech and UCSD) to the common stereotaxic coordinate system, and converted them into ArcIMS image services capable of generating image fragments at appropriate resolution on client requests. Segmented and topologically-cleaned anatomic feature data are served via ArcIMS feature services. Additionally, large brain images placed under control of SRB (Storage Resource Broker), the data grid environment adopted by BIRN, can be queried via an SRB proxy responsible for image cutting and downsampling given the spatial extent of the requested fragment. All image and feature services are registered at a spatial index source which includes, for each image or feature slice, its coordinate system, the plane it belongs to, and location of the slice within the plane (using standard GIS mechanism of "world files".) With this information, users can answer queries such as "find brain data sources that contain images and markup for a user-defined area of the brain". Once discovered, the images and segmented anatomic feature can be retrieved, overlaid, and further queried in the client Smart Atlas application. For example, it is possible to display image segmentations available in one atlas source, over image fragments retrieved from a different atlas source, or compare protein distributions in a given area of the brain across atlas sources. The comprehensive integrated atlas environment opens new opportunities for scientific discovery in neuroscience.