Image BOSS: an image database system designed for research.

Abstract

T HE PRIMARY PURPOSE for the use of a database is to help manage the vast amounts of data generated in today's information age. Like other types of data, biomedical image data is currently being generated at enormous rates, and shows no signs of decreasing. Large volumes of image data become unmanageable under current computational systems. Mayo Foundation supports a significant number of research studies, which include medical images (computed tomography, magnetic resonance, etc) and biological images (optical microscope, confocal microscope, etc). Research topics range from basic science through clinical diseases. The Biomedical Imaging Resource (BIR) is a core resource at Mayo with a mission to provide expertise and computer resources for advanced image visualization and analysis. The BIR affiliated research activities generate large volumes of image data. For example, gastrointestinal nerve studies, which derive images from optical microscopy typically generate 50 to 70 images per nerve, each image containing 512 • 512 8 bit pixels, or approximately 262 kilobytes each. Similar cell types are imaged using different staining techniques. Several sets of images are collected from different organs in the body, and several different types of animals are under study, resulting in hundreds of sets of raw image data. After the raw data is collected, image processing is used to register, segment, and scale the images into an isotropic volume image. From these volume images, measurements are taken, objects are segmented and various visualization techniques, such as volume and surface rendering generate further images, which are used to view the cells and study the 3-dimensional relationships in the data. To date, the BIR has accumulated over 5,500 images for this single physiology investigator's work, requiring over 40 Gigabytes of storage.