A Vanderbilt Discovery Grant is funding David Piston’s plan for high volume image data storage, management and processing..

Servers, database improve image access

During bursts of dynamic imaging, the microscopes in Vanderbilt’s Cell Imaging Shared Resource can generate data at a rate of more than 240 megabytes per minute, says David Piston, the core’s scientific director. That’s about one CD-ROM’s worth of images every three minutes, he says.

Where do you store all of these data, and how do you access and analyze the images efficiently? These are the kinds of questions being addressed by imaging bioinformatics, Piston says. He was recently awarded a Vanderbilt Discovery Grant to implement a plan for high volume image data storage, management, and processing.

Piston’s plan involves creating a centralized repository for storing images and structuring that repository so that access is efficient. “Simply having a large amount of disk space doesn’t guarantee access to it in a timely fashion,” he says.

Traditionally, data have been stored on individual disk servers. The problem with this scheme, Piston says, is that an individual – during large file transfer – can monopolize a single server and the network, leaving other disk servers sitting idle. To overcome this problem, Piston will generate a parallel file system – essentially a group of servers transferring data in parallel, instead of independently.

To manage the stored images for analysis and sharing with other investigators, Piston will implement a database scheme developed by the Open Microscopy Environment (OME, http://www.openmicroscopy.org). OME is an open source software project to develop a database-driven system for quantitative analysis of biological images, Piston says. It is a collaborative effort among academic and industrial labs that was started in 2000. The first versions of OME programs – standardized file formats for image data and database schema – are nearing completion.

Having an image database will allow researchers to systematically deposit data and selectively extract images from simple queries to the database. Keyword searches or other cross-referencing techniques will provide investigators with the opportunity to compare images from different laboratories and experiments, Piston says.

Piston has teamed up with Alan Tackett at Information Technology Services to install, maintain and back up the parallel servers. “In addition to using existing expertise at ITS, this arrangement will allow much wider access to these services once we get them established,” Piston says. The server and database system will benefit many different groups, including users of the Cell Imaging Shared Resource, the In Vivo Imaging Center, which Piston also directs, and the newly created Vanderbilt Institute of Imaging Science under the leadership of John Gore.

“We expect this server and database system to greatly advance image informatics and to accelerate progress in biological and biomedical research at Vanderbilt,” Piston says.

In addition to providing user-friendly storage and access to image data, Piston expects the new system to speed data analysis. The OME database will interface with both existing commercial image processing software and with new analysis routines.

Piston and colleagues are developing one such new routine for specialized particle analysis. They are using so-called deformable models to track moving subcellular particles – such as mitochondria and insulin granules, which they image to study glucose-stimulated insulin secretion – over the time course of an experiment. The new semi-automatic and automatic methods will allow analysis, processing, and visualization of large three- and four-dimensional data sets.

“The server and database system will make these kinds of analyses possible,” Piston says.