Cellular responses are heterogeneous, tissue specific, and a function of the history of a cell and its genome. In dealing with the heterogeneity of multiple model systems plus in-vivo studies, each proposed project will generate a large number of specimens for detailed quantitative and correlative analyses. The Imaging Bioinformatics Core will complement and extend the presently developed BioSig framework with two objectives: (1) to provide a fully annotated set of representative samples that are imaged at different resolutions, and (2) to populate databases that link anonymous patient data to mammography, breast density and expression profile data plus data obtained from histological analyses. For this objective annotation refers to user's input and feature-based representations that are computed using image analysis techniques. The first goal will target Projects 2, 3, and 4, and the second goal will target all Projects and Cores. Detailed quantitative representation of data enables comparative analysis of images based on their content, while linking data from different modalities enables event correlation and information visualization. Quantitative representation will be applied to (1) low-resolution compositional analysis of breast density, (2) low-resolution 3D modeling of ductal tree structures from regions of high and low breast density, (3) high-resolution 2D and 3D morphological and protein localization studies, and (4) analysis of expression profiles in support of Project 2. Compositional analysis will investigate the ratio of epithelial, stroma and adipose in low- and high-density regions. 3D representation of ductal tree structures enables comparative morphological analysis between different regions of breast tissue and quantitative analysis of high-resolution image data enables morphological and protein expression analysis using markers that target specific inter- and intracellular activities in tissue or cultured multicellular systems. The Core will couple user-defined annotations with the raw data and their computed annotations to (1) enable navigation between different data modalities, (2) provide graph-based queries, and (3) view the results through a Web-based interface in the form of plots, scatter diagrams, or images. This core enables sharing of data with collaborating investigators outside of the program project. The core will leverage the BioSig framework (developed at LBNL) and GeneTraffic platform (developed at lobion) in support of analysis of images through microscopy and microarray studies. The Core will extend the current ontology for managing radiological data, construct 3D models of the breast from Egan slices, and develop software tools to overlay gene expression and patterns of protein expression onto this 3D space for meaningful information visualization. The Core will enable navigation and query of this heterogeneous data space through graphical model, common schema, and controlled vocabulary. Quantitative representation of images and their annotation will be accessible to the BioStatistics Core for detailed sensitivity analysis.
Showing the most recent 10 out of 19 publications
