Image is Everything

From 2-D electrophoresis to microscopy, much of the data generated in today's laboratories is image-based. Until recently however, there was no way to efficiently search the contents of images and to correlate them with other types of data. To address this problem, two companies have developed systems that bring the full power of the informatics revolution to bear on visual data. According to Suzanne Mattingly, vice president of marketing at Redwood Shores, Calif.-based Scimagix, up to 70 percent of R&D data exists in the form of images, with much of it stored in disparate, removable storage media that is often difficult to access and use. Scimagix's first commercially available product, the Oracle®/Web-based SIMS™ (Scientific Image Management System), is a database package for the storage, analysis, and mining of annotated experimental image data. SIMS is not an image acquisition product--images are scanned in and stored in the database, along with other relevant data. Thus, the software enables users to tie together image data, as well as proteomics, gene expression, and preclinical data, across the various stages of a given project. This integrated data can therefore be used more effectively by different groups of researchers involved in the development of a particular drug, for example. Robert Dunkle, president and CEO of Scimagix, observes that integrating image data with other types of information to produce "image-driven structure/activity relationships" (ISAR), will ultimately improve researchers' ability to understand the mechanisms studied, regardless of where and how the images are generated. As Dunkle explains, "We have the ability to integrate images with chemical and biological data--for example--so you can look at the total picture of what's happening to gain insights and make better decisions about which compounds to progress -- or not. By relating proteomics images to gene expression data and even preclinical data, a more exacting picture comes into play. That hasn't really been possible before." Any information that the user defines as important--for example, drug type, dosage, or outcome--can be stored along with image data to facilitate searching. Courtesy of ScimagixScimagix's SIMS database SIMS is a highly versatile application. For example, Scimagix has collaborated with Parke-Davis, a division of Pfizer of Groton, Conn., to develop a visual search application to recognize and search protein patterns in 2-D electrophoresis gels. This approach will soon be applied to cellular and tissue assays and enable researchers to search by parameters based on color, size, shape, and texture. Additionally, researchers at Eli Lilly & Co. of Indianapolis currently use Scimagix's software to integrate tissue images from clinical trials and gene expression data from discovery research. While the Scimagix system facilitates storage and retrieval of image-based information from a variety of sources, a new product from TissueInformatics Inc. of Pittsburgh, Pa., focuses specifically on tissue data. The Tissue Information Solution™, based on the company's Quantf(x)™ technology, captures, analyzes, searches, and mines microscopic tissue image data and automatically quantifies any user-defined feature related to a particular image. "The tools we develop are imaging tools, so that you can actually see the things in tissue that you want to see, [and] land mapping tools for tissue-automated software that analyzes tissue, locates all the components within it, including cells, matrix, and structures, and then identifies their coordinates within the data set," says Peter Johnson, president and CEO. The information collected and archived by the system can then be used for drug discovery and tissue engineering applications. Dermf(x), the company's first release, analyzes the structure of human and engineered skin. "The program identifies the tissue against a background, locates all of the cells within the tissue, determines the thickness of all layers, and locates anything else of interest that has been programmed into the software. The data is then extracted as a set of x, y coordinates into a binary data set that can then be mined for quantitative information about any of the elements," says Johnson. This information can then be correlated with other sets of data collected at the same time, including clinical, genetic, and proteomic information. Currently under development are applications for high-throughput antibody screening, in situ hybridization, and blood vessel analysis. Additionally, the company develops quality assurance and tissue structural mapping capabilities for the tissue engineering industry. Mark Braughler, senior vice president for business development, estimates that TissueInformatics' customers are divided evenly between pharmaceutical companies and tissue engineering firms. While the data generated and stored by the software is valuable in itself, the ability to integrate this data with other types of genomic and proteomic information makes the technology created by these two companies central to the field of bioinformatics. Although image may not always be everything, image informatics is likely to enhance the process of drug discovery by bringing together the right information at critical decision points. -Aileen Constans (aconstans@the-scientist.com) For More Information Scimagix Inc. (650) 508-2270 www.scimagix.com TissueInformatics Inc. (412) 488-1100 www.tissueinformatics.com

Image is Everything

Interested in reading more?

Become a Member of