Micro-To-Macro Multimodality Atlas Formation

点击翻译按钮获取中文摘要

DESCRIPTION (provided by applicant): Quantum leaps in medicine and the biological sciences will be achieved when anatomical and functional details at multiple scales can be linked within and across subjects. The technology for capturing, managing, and reporting such details and linkages is the science of atlas formation. While images have the potential to capture micro and macro details, an atlas extends simple image representations to include relationships between data at different scales, further supplemented with annotations, web content, and animations. An atlas also provides an ordering to the information so that consumers can readily comprehend the atlas content. We propose to develop software applications that enable researchers with limited software and image processing expertise to create and interactively visualize multi-scale, multimodality, image-based atlases. This work builds upon industry-standard software libraries that we (the CADDLab at the University of North Carolina and Kitware, Inc.) were instrumental in developing: the NLM's Insight Toolkit (ITK) for medical image analysis and the Visualization Toolkit (VTK) for data visualization. To demonstrate the functionality of our software applications, pathologists will apply them to link confocal microscopy images of histology with micro-CT and MR images to illustrate the association between heart muscle gene expression patterns and cardiac hypertrophy in mice. The specific aims of this Phase I application are as follows: 1. To provide validated software methods for stitching and stacking multiple confocal microscopy images into a 3D representation of a specimen. 2. To design and demonstrate a pluggable software framework for image-based atlas formation supporting the tasks of registering, segmenting, annotating, and linking histological and in vivo images. 3. To extend an existing visualization product to support atlas viewing: visualizing large images, managing annotations, and traversing links to external content.