NeuroBridge ontology development for shared neuroimaging datasets
Presenting author:
The central problem in reusing data from published studies is finding these datasets. Despite the introduction of public data repositories, there are many publications where the data is not archived or only "available upon request." Our goal is to develop methods for searching the literature for available datasets from projects of interest.
We have developed the NeuroBridge ontology (NBO) for representing the types of data collected in an experiment: subject population, imaging techniques used, and relevant behavioral, cognitive, or clinical data. The NBO uses the metadata standard called W3C PROV and extends the ProvCaRe ontology that was developed to support scientific reproducibility. It reuses terms from NIDM, SchizConnect, and SNOMED CT to model metadata describing study data, study methods, and study instruments. The NBO currently includes 623 terms.
We have developed and tested the NBO terms by hand annotating a corpus of recent papers describing imaging in schizophrenia studies and substance abuse (180 papers from 2018-2021). These have been annotated with the NBO terms described above.
We are extending the NBO with more information about the different tasks included in the functional neuroimaging, as well as expanding the NBO’s ontology for clinical, cognitive, and behavioral data; incorporating terms from other ontologies. The next stage of this work will apply this ontology to automatically annotate additional papers and mediate queries over research datasets.
We have developed the NeuroBridge ontology (NBO) for representing the types of data collected in an experiment: subject population, imaging techniques used, and relevant behavioral, cognitive, or clinical data. The NBO uses the metadata standard called W3C PROV and extends the ProvCaRe ontology that was developed to support scientific reproducibility. It reuses terms from NIDM, SchizConnect, and SNOMED CT to model metadata describing study data, study methods, and study instruments. The NBO currently includes 623 terms.
We have developed and tested the NBO terms by hand annotating a corpus of recent papers describing imaging in schizophrenia studies and substance abuse (180 papers from 2018-2021). These have been annotated with the NBO terms described above.
We are extending the NBO with more information about the different tasks included in the functional neuroimaging, as well as expanding the NBO’s ontology for clinical, cognitive, and behavioral data; incorporating terms from other ontologies. The next stage of this work will apply this ontology to automatically annotate additional papers and mediate queries over research datasets.