Abstract
Abstract: :
Purpose: To present the LSU Eye Center Media Repository System, a hardware/software solution built for the storage, search, retrieval, and display of large data sets. The system supports heterogeneous media and data types and also allows immersive visualization of three–dimensional data. Methods: Imaging has become an essential tool in bio–medical research. Laboratories are increasingly using high–resolution, real–time digital imaging and video. These advances come at a price: the amount of information being generated is growing at a tremendous rate and, therefore, storage, search and retrieval, and visualization systems are needed for the efficient management and analysis of these data. These systems typically involve integration of high–performance networks, hardware, and software components, as does our solution. A 4–way IBM p650 server running IBM DB2 database and WebSphere application server handles all data requests. Data are stored in an IBM NAS 25T storage array with 4 TB raw capacity (1.8 TB fault–tolerant), and three–dimensional data are visualized in a Fakespace Immersadesk M1 immersive display driven by a Sun Fire V880z visualization server. The system accepts media and data of any type and currently supports the management and visualization of audio/video, static images, image stacks (volumes), and contoured surfaces. Server–side and client–side software is written in Java for portability. The software supports intranet and remote access from heterogeneous client machines. Results: The system is still under development and has been deployed on a limited basis for trial purposes. Initial tests included the storage, 3D reconstruction, and immersive visualization of image stacks from the lamina cribrosa of the cow eye. Legacy data are being migrated to the central database as time and resources allow. Efficiency has been greatly improved given that all data now sit on a central repository and a single client browser can retrieve these data with visualization on different devices. Conclusions: Centralized data storage has many benefits: performance, fault tolerance, increased availability, and efficient searches among heterogeneous data types. Initial conversion to such systems is initially difficult and time consuming but the benefits far outweigh the costs.
Keywords: computational modeling • image processing • imaging/image analysis: non–clinical