Connecting Brain Imaging Acquisition Protocol, Processing and Analysis Terms to an Established Lexicon
Massachusetts General Hospital, USA
Massachusetts Institute of Technology, USA
University of Washington, Department of Medical Education and Biomedical and Health Informatics, USA
University of California, Irvine, Department of Psychiatry and Human Behavior, USA
University of Warwick, Warwick Manufacturing Group and Dept. of Statistics, United Kingdom
The MIND Research Network, USA
For data sharing to be useful, data must not only be easily available and stored in an organized fashion, but the metadata that captures information about how the data was acquired, processed, and analyzed, must also be available. Additionally, metadata must describe this data using unambiguously defined terms. Efforts are underway to provide lexicons of brain imaging terms to the neuroscience community. Two examples are NeuroLex and RadLex; lexicons for the domains of neuroscience and radiology, respectively. While NeuroLex follows the OBO principle of always defining a term, RadLex identifies relationships among terms, but often without definitions.
Often the item to be shared resides in data or tool repositories, which may have a defined schema or a fixed set of terms, but do not provide definitions for those terms. Users may not know precisely what is meant by each term and this makes it difficult to meaningfully combine data from disparate sources. In addition, the lack of standardized terms makes it difficult for query tools to search across collections at different institutions and makes data provenance ambiguous.
The overall goal of this project is to provide definitions for terms used in each stage of the data lifecycle of brain MRI-based experiments and place the terms within NeuroLex. We have begun our efforts with DICOM  terms, neuroimaging data processing and analysis terms, and BIRN’s Human Imaging Database (HID) terms. In some cases, connecting or parent terms were added to Neurolex to connect the new set of terms with those already in the lexicons.
To date, we have added over 1700 DICOM tag terms to Neurolex. We are currently evaluating systems that allow the creation of Terse RDF Triple Language (Turtle) files of these terms and their relations for importation into other ontologies or lexicons. Work is also ongoing to structure the data processing and analysis terms for addition into Neurolex. This work expands an existing lexicon with terms commonly used in neuroimaging and their inclusion will allow formal data models to refer to these terms through a SPARQL endpoint.
This work was conducted within the Derived Data Task Force of the International Neuroinformatics Coordinating Facility Program on Standards for Datasharing. KGH, DK, and JT are supported by Biomedical Informatics Research Network (BIRN) through 1U24-RR025736.
1)http://neurolex.org, 2)http://www.radlex.org, 3)http://www.obofoundry.org 4)http://medical.nema.org/, 5)http://www.birncommunity.org/
Neuroinformatics 2013, Stockholm, Sweden, 27 Aug - 29 Aug, 2013.
(2013). Connecting Brain Imaging Acquisition Protocol, Processing and Analysis Terms to an Established Lexicon.
30 Apr 2013;
11 Jul 2013.
Dr. Karl Helmer, Massachusetts General Hospital, Charlestown, Massachusetts, 02129, USA, email@example.com