Notice: Trying to access array offset on value of type null in /home/arlaird/ on line 99

Notice: Trying to access array offset on value of type null in /home/arlaird/ on line 99

Notice: Trying to access array offset on value of type null in /home/arlaird/ on line 100

Notice: Trying to access array offset on value of type null in /home/arlaird/ on line 100

Notice: Trying to access array offset on value of type null in /home/arlaird/ on line 101

Notice: Trying to access array offset on value of type null in /home/arlaird/ on line 101
Aims of the CogPO Project - CogPOwiki

Aims of the CogPO Project

From CogPOwiki
Revision as of 14:14, 3 August 2011 by Laird (talk | contribs) (Undo revision 1200 by Zhangde (Talk))
(diff) ← Older revision | Latest revision (diff) | Newer revision → (diff)
Jump to: navigation, search

A. SPECIFIC AIMS. The objective of this proposal is to develop, evaluate, and distribute a domain ontology of cognitive paradigms for application and use in the functional neuroimaging community. This cognitive paradigm ontology, CogPO, will be developed through the integration of two well known and established databases, the Functional Imaging Biomedical Informatics Research Network (FBIRN) Human Imaging Database (HID) and the BrainMap database.

Neuroimaging research has produced an enormous amount of data suitable for mining, analysis, and meta-analysis. As a byproduct of this, new databases are constantly emerging. Yet until a complete ontology is established and adopted, communication and the exchange of information within and between these neuroimaging databases will be limited. There are numerous of existing efforts that aim to develop ontologies: RadLex, an ontology of medical imaging acquisition strategies; lexicons and ontologies of neuroanatomical regions (e.g., NeuroNames and the Foundational Model of Anatomy (FMA)); full medical ontologies for clinical care concepts, such as the Systematized Nomenclature of Medicine (SNOMED) Ontology, and many others. In addition, NeuroLex, pulls from all these in developing ontologies of neuroimaging methods and datasets.

Despite the number of ontologies currently being developed, no ontology effort exists to formally define and characterize the cognitive paradigms that are used in functional neuroimaging studies. Even a brief examination of the literature reveals that a given cognitive paradigm may be encountered under a variety of different names. For example, the “Sternberg task” may also be identified as the “serial item recognition paradigm (SIRP)”, a “delayed match to sample task”, or merely as a “working memory task”. The latter example is especially confusing as it conflates what the paradigm is intended to study at the time of use, with what it might be understood to characterize later. That is, applying the label of “working memory task” to data acquired during performance of the “n-back task” presupposes that the data would not be useful in examining “executive function”. This frequent use of alternate and sometimes competitive terminology makes the formal development of an ontology based solely on paradigm names extremely difficult, if not impossible.

The design of CogPO concentrates on what can be observed directly: categorization of each paradigm in terms of (1) the stimulus presented to the subjects, (2) the requested instructions, and (3) the returned response. All paradigms are essentially comprised of these three orthogonal components, and formalizing an ontology around them is a clear and direct approach to describing paradigms. This type of classification has been a core feature of the BrainMap database since its inception, was endorsed and refined by many reviewers during previous BrainMap workshops (1994-1998), and was part of the BrainMap ontology presentation to the Neuroimaging Informatics Technology Initiative (NIfTI). Categorization of paradigms according to stimulus, response, and instruction has been shown to allow advanced data retrieval techniques by searching for similarities and contrasts across multiple paradigm levels. This approach is less susceptible to the whims of conceptual shifting, in that while one’s research interests may progress from “working memory” to “executive function”, one may continue to be interested in paradigms in which (1) a target set of letters were presented, followed by a probe letter, (2) after which subjects indicated if the probe letter matched any target, (3) using a button press response.

Rather than develop a cognitive paradigm ontology in the abstract, the aim of this proposal is to develop and evaluate CogPO in the context of two neuroimaging databases, the FBIRN Human Imaging Database (HID) and the BrainMap database. The FBIRN HID and BrainMap serve very different missions. While the FBIRN HID aims to archive image data prior to publication for re-analysis and mega-analysis, BrainMap’s goal is to archive reduced (coordinate) data for meta-analysis. Because of these fundamental differences, FBIRN HID contains a smaller number of highly focused and finely detailed paradigms, while BrainMap contains a very large number of paradigms represented at a coarser resolution. The present proposal seeks to take advantage of these differences by creating an iterative model of ontological development.

Aim 1. Develop the core ontology for the common paradigms in BrainMap and the FBIRN HID. Using the example cognitive paradigms available in both data repositories, and building on our previous work already incorporated into BIRNLex and the BrainMap coding scheme, we will develop and implement the core ontology to fully capture the paradigm characteristics, which are currently contained in both BrainMap and the FBIRN HID, but not ontologically mapped. This will involve defining the hierarchical relationships and instances of stimuli, responses, and instructions in a machine-interpretable and publishable format.

Aim 2. Extend the ontological mapping to the complete list of paradigms contained in BrainMap. We will extend the ontological mapping procedure developed and refined during Aim 1 to encompass all of the paradigms contained in BrainMap. Inclusion of the total list of BrainMap paradigms is an additional, indirect method of involving the community in the development of this ontology, as BrainMap is an open database of neuroimaging results and is therefore potentially represented by any member of the community.

Aim 3. Enable information exchange between FBIRN HID and BrainMap. The true value of a given ontology can be measured in its ability to facilitate data retrieval. To this end, we will test the efficacy of the ontology developed in Aims 1 and 2 by enabling interoperability and the exchange of information between both databases according to cognitive paradigms of interest. Through combinations of CogPO and other ontologies, the possibility of automated hypothesis testing will be explored.

Sharing Plan. The present proposal seeks to take advantage of ontological best practices that have shown to be valuable in other projects. Through a series of workshops, we will consult both the domain experts (the neuroimaging research community) and the ontological experts (the National Center for Biomedical Ontology (NCBO) and the BIRN Ontology Task Force (OTF)) as the ontology develops. We will also allow user input and feedback from the entire neuroimaging community through the Neuroimaging Informatics Tools and Resources Clearinghouse (NITRC) and the NCBO BioPortal, with links to a Wiki-based presentation of the concepts similar to that used by the Neuroscience Information Framework (NIF) or the Gene Ontology Consortium. At the end of funding, we aim to deliver a complete and functioning ontology of cognitive paradigms used in functional neuroimaging in the Protégé-OWL format. These files will contain the concepts, definitions, and relationships of paradigms represented by stimuli, responses, and instructions, and will be distributed through NITRC and the NCBO BioPortal. This ontology will be made available for adoption not only by other fMRI databases, but also for archives of other neuroimaging modalities (e.g., EEG or MEG data), such as the Neural ElectroMagnetic Ontologies (NEMO), and literature neuroinformatics efforts such as the Society for Neuroscience’s PubMed Plus.