Neuroinformatics is a branch of biomedical informatics. The field is a heterogeneous mixture of neuroscience, cognitive psychology, information science, computer science, database science and clinical research informatics. Neuroinformatics predominately deals with the storage, retrieval and organization of neuroimaging data from clinical research, as well as nonclinical basic research.
Current neuroinformatics research involves:
1) Tool and database development
2) Novel analysis methods for big data within neuroimaging
3) Advanced computational neural models
Examples include: designing a tool or suite supporting automated metadata capture, recording data provenance, tools for automating data conversions, standard data acquisition protocols for reduction in variability between datasets, standard data sharing protocols, and novel image retrieval techniques for secondary image analysis and pattern recognition.
There are various neuroimaging modalities which include: magnetic resonance imaging (MRI), functional magnetic resonance imaging (fMRI), transcranial magnetic stimulation (TMS), electroencephalography (EEG), and magnetoencephalography (MEG). The various imaging modalities may include a combination of imaging sequences within a modality (e.g. resting state fMRI versus active fMRI). The different neuroimaging technologies capture different types of brain activity data from surface cortical activity to morphological changes and variations in blood flow. The most abundant and commonly used imaging modality is MRI and fMRI. Despite differences in imaging modality, most neuroimaging studies capitalize on the DICOM standard interoperability between devices (tablet, computer, imaging source).
The DICOM standard has been used for distributing and viewing any kind of medical image regardless of origin. The DICOM standard specifies a standard file syntax and metadata structure, along with a standard protocol for transferring images among devices. Since 1995, all major diagnostic imaging modalities have been standardized to DICOM (Bidgood, et al., 1997); however, not all neuroimaging parameters have been standardized to DICOM. For example, there are certain parameters for fMRI, diffusion tension imaging, chemical shift spectroscopy and perfusion-weighted imaging that require more header fields not available in DICOM. There are other file formats for neuroimaging files such as ANALYZE 7.5, ECAT, GE, MINC, NIFTI and NRRD file (Neu, Crawford, & Toga, 2012).
Neuroimaging databases are quickly proliferating, however, there is wide variation in the type of databases being created. In the neuroscience community, the scope of the database often falls in one of three different categories: 1) A database for a single laboratory, 2) A collaborative database for a research community, 3) A public database. The National Institutes of Health launched an initiative called the Biomedical Informatics Research Network (BIRN) to share and mine data from both basic and clinical research. The BIRN project uses a series of distributed databases by which data are shared and queried at participating institutions. The databases were designed to span a variety of technologies and species. Although each database is designed to hold different information, it is also designed to be interoperable and applicable across studies. The BIRN architecture is based on relational XML databases. Another Neuroinformatics databases is the Laboratory of Neuro Imaging (LONI) at the University of California Los Angeles an XML schema document (XSD) is used to define the elements and attributes of a data type (MacKenzie-Graham, et al., 2008).
A particular hurdle to database interoperability is the creation of a standard neuroimaging vocabulary. The University of California Davis has created an ontological server to maintain a neuroscience thesaurus combining concept identifiers with other medical thesauri such as Systematized Nomenclature of Medicine (SNOMED). The thesaurus also provides indexing information for image annotation, a central metadata repository and information on spatial and temporal relationships. Among neuroscience anatomy terms, spatial relationships link spatial regions of interest. Temporal relationships map time-dependent functional activity changes in spatial regions of interest. Information in XML format, including semantic metadata, can be integrated with metadata for relational or object-oriented databases.
XML is an emerging standard for data description. In XML there is a hierarchy of tags, nested tags, and schemas. Most neuroimaging information including datasets, metadata, queries, algorithms, and transforms can all be defined and transferred via XML tags (Bowden & Dubach, 2005).
Bowden, D. M., & Dubach, M. (2005). Neuroanatomical nomenclature and ontology. In S. H. Koslow & S. Subramaniam (Eds.), Databasing the brain: from data to knowledge (neuroinformatics) (pp. 27-46). Hoboken, New Jersey: John Wiley & Sons, Inc.
Gorin, F., Hogarth, M., & Gertz, M. (2001). The challenges and rewards of integrating diverse neuroscience information. Neuroscientist, 7(1), 18-27.
MacKenzie-Graham, A. J., Van Horn, J. D., Woods, R. P., Crawford, K. L., & Toga, A. W. (2008). Provenance in neuroimaging. Neuroimage, 42(1), 178-195. doi: 10.1016/j.neuroimage.2008.04.186
Neu, S. C., Crawford, K. L., & Toga, A. W. (2012). Practical management of heterogeneous neuroimaging metadata by global neuroimaging data repositories. Front Neuroinform, 6, 8. doi: 10.3389/fninf.2012.00008
Submitted by (F Cook)