Content-Based Indexing and Retrieval of Brain Odor Maps

点击翻译按钮获取中文摘要

DESCRIPTION (provided by applicant): Neural signals invoked by odor stimuli are encoded in space in the mammalian olfactory bulb (OB). Functional magnetic image (MRI) odor maps are single two-dimensional flat brain images that describe the spatial activity patterns in the entire OB glomerular layer in response to a given odor stimulation. The MRI odor maps and annotations, such as odor type or exposure duration, are saved to the database and can be retrieved based on the annotating text. However, the availability of search tools based on the map content becomes increasingly important as more maps are deposited and the activity patterns of the maps become more complex. The lack of such tools is evidently hindering the full usage of any odor map database. The aim of this proposal is to design and implement a database system with informatics tools for the content-based indexing, storage, and retrieve of the MRI odor maps. The system will be built with Java and web technologies and on the foundation of the current MRI odor map database. Based on the previous knowledge and our recent studies, the proposed project will create sets of templates for the rodent MRI odor maps. The templates will include registered modules. Both the templates and modules are stored in the database. The new tools will allow users to perform image warping and align maps to the appropriate template. The system will then extract the pixel information from each module. For odor maps submitted for archiving, the indexed information will be saved to the database. The proposed web interfaces and tools will allow users to search maps based on anatomical content or using an unknown query map which will also be subject to the indexing process. Similar tools with advanced features including threshold-dependent and Boolean search will also be designed and implemented. For instance, given an unknown query map with certain activity patterns, researchers will be able to retrieve all the maps with similar patterns, the maps including or excluding those patterns. The Boolean search will play an important role in analyzing the complex odor maps generated by a combination of different odor types. The informatics tools developed under this proposal will facilitate the experimental MRI research in the field, help to discover the unique nature in olfactory signal encoding in the mammalian olfactory system, and to understand the mechanism of odor perception.