Speaker of Workshop 1
Will talk about: Essential features of digital brain models for neuroinformatics
Douglas Bowden is the creator of BrainInfo, a portal to neuroanatomical information on the Web; NeuroNames, the ontology of neural structures on which BrainInfo is based; and NeuroMaps, the topic of his presentation at this meeting. NeuroMaps is a web resource that enables researchers to map data to standard atlases of mouse and macaque brains and to prepare images for presentation and publication. BrainInfo links visitors to informative pages at some 100 other neuroscience websites. It provides basic information on some 3000 central nervous system structures, which can be accessed by 18,000 neuroanatomical terms in eight languages. It is used throughout the world by some 400 visitors per day. NeuroNames has become a source vocabulary of the Unified Medical Language System (UMLS) of the National Library of Medicine. It serves as the basis for neuroanatomical ontologies of the Foundational Model of Anatomy, NIFSTD of the Neuroscience Informatics Framework (NIF), and a number of individual laboratories that have adopted the vocabulary for indexing databases of neural information. Dr. Bowden is Professor of Psychiatry and Behavioral Sciences and Core Staff Scientist at the National Primate Research Center at the University of Washington, Seattle. BrainInfo’s sponsors include the University of Washington, the NIF, and the International Neuroinformatics Coordinating Facility (INCF).
The ideal digital brain atlas will allow investigators to map data obtained by any method directly into a canonical atlas where its location can be compared with all other data mapped to the same atlas. The neuroscience community will be best served by adoption of a single atlas per species, because every transcription of data from one atlas to another loses information. For maximal precision the atlas of each species should be based on a high resolution MRI. To maximize utility for neurophysiologists who wish to stimulate, inject, or record in areas of data concentration, the atlas should be registered to a conventional stereotaxic space. Since every investigator’s first question will be, “Where are my data located?” the canonical atlas should be segmented to match a widely accepted conventional atlas. Alternative segmentations should be mapped to the atlas like any other data and made available for comparison. The mapping application should enable nonlinear warping of images of brain sections to corresponding planes of section in the atlas using as many landmarks as possible. The atlas application should enable preparation of images of mapped data for presentation and publication. And it should enable investigators to upload mapped data sets to the atlas website: 1) for comparison with existing data, and 2) for deposit in a permanent repository where future investigators can compare it with theirs.