Mbirn: Ontology: Project Background

• The overarching goal is to create an ontology prototype that relates brain structure and function through neuroanatomical regions, neuropsychological and cognitive terms, and clinical assessments.
• The BIRN’s Mediator tool will provide an integrated view of the BIRN shared database infrastructure. That is, the Mediator will allow the user to “pretend” that the information from a variety of available databases is from the “same” database. The user should be able to request results through a query of all available assessment scores (e.g., from a memory test) across sites and pull this same data seamlessly.
  • To do this, the Mediator relies in part on a defined Ontology structure to define the relationship(s) between similar terms.
  • For example, if a user is interested in finding data related to "Purkinje cell," it would be important to also include information related to "Purkinje neuron." This type of relational knowledge is critical. The ontology will be used to find other values in other records in other databases that have similar information.
  • In addition, the ontology will provide a semantic network; for example, for a user searching for information on "cerebellum" information related to the cerebellum would include both Purkinje concepts previously mentioned as well as additional information.
• The current MBIRN testbed goal is to extend clinical research examples from the Multi-site AD project to a relationship-based ontology extended from the UMLS (Unified Medical Language System, U.S. National Library of Medicine/NIH http://www.nlm.nih.gov/research/umls/index.html ). The UMLS provides a vocabulary database stored in a “Semantic Network” that provides categorization of concepts and relationships between concepts.
• Our preliminary focus is a basic neuropsychological ontology as it relates to clinical assessments, cognitive function, and neuroanatomy in the Multi-site AD sample, with future branching into neuropsychiatric measures (e.g., fBIRN schizophrenia interviews, etc.). While extremely vast, the UMLS does not provide all concepts or all types of relationships that will be valuable to our clinical imaging research questions.
  • For example, the UMLS contains a concept for "California Verbal Learning Test" which is a neuropsychological assessment of learning and memory. It does not, however, contain the concept "CVLT" which is the common acronym for the assessment; this concept will need to be added.
  • More importantly, this concept only has defined relationships with the concept "Assessment Scales" and to other assessments scales in UMLS; there is no defined relationship between this measure and either "learning" or "memory." This latter development of relationships between clinical assessments, cognitive function, and, ultimately, to neuroanatomical terms (e.g., the hippocampus) is the focus of the current project.
• To create extensions and additions beyond the existing NLM ULMS without direct modification, the BIRN has developed "BONFIRE," a method for creating a transparent extension to the UMLS; BONFIRE is a separate database of new information in the same structure as UMLS and provides similar information on meaning and relationships. BONFIRE will maintain all new extensions to UMLS, and users will be able to view and graph information just as in UMLS. For a view of the existing BIRN tool visit http://nbirn.net/ and login to the portal; then select Advanced Data Integration and KNOW-ME UMLS ( http://imhotep.ucsd.edu:7873/knowme/umls.html ).
• Additional methods for data presentation, including graphical interface and semantic organization structures, are under development. In particular, work with David Karger's group (MIT) will provide an interface between the BIRN interface and Haystack, a semantic organization tools, for future data viewing.

For more information see:

• Ontology: Description of Development Plan
• Ontology Updates

---