Image Analysis Under Windows: New Tools For Biologists

At root, image analysis is the ability to convert a picture into hard data. Over the past few years, a family of versatile, easy- to-use software packages has emerged that promises to put that ability into the hands of more life sciences researchers than ever before. "If a picture's worth a thousand words, a number's worth a thousand pictures," says Paul Goodwin, explaining the technology's value to scientists. Goodwin is manager of the image analysis laboratory of the Fred Hutchinson Cancer Research Center, Seattle. "Historically, biology's been very descriptive. The problem is, the more descriptive a subject is, the more subjective it is. Our goal [in image analysis] has been to find some way to objectify those descriptions." Taken together, these new image-analysis systems take advantage of the fast-dropping prices and growing computational power of IBM-compatible personal computers (PCs). They also build on the strengths of the Windows operating system, the more intuitive graphical user interface (GUI) that progressively is replacing the MS-DOS operating system on many scientific desktops. Among these image-analysis software packages are: OPTIMAS from BioScan Inc., Edmonds, Wash.; GLOBAL LAB Image from Data Translation Inc., Marlboro, Mass.; and MOCHA from Jandel Scientific, San Rafael, Calif. In addition, Media Cybernetics in Silver Spring, Md., expects to release a Windows version of its product, Image-Pro Plus, in June, according to marketing manager Cynthia A. Batz. "A lot of our current users have asked for Windows," Batz says. "So, they're obviously porting or moving over to that platform." According to Tom Bligh, president of BioScan, there are five basic steps to the image-analysis process incorporated into these software packages. The first step is to capture the image in a digital format usable by the computer. Formerly, an expensive piece of hardware called a frame grabber was required for this step, and for many uses this is still the case. But desktop publishing scanners also can be used to scan gel electrophoresis results, for instance, directly into the computer. The second step is to enhance the image by adjusting contrast thresholds, for example, or by averaging several images to reduce instrument noise. Identifying which objects in the image the analysis will focus on is the third step. This can be done manually or automatically, with the help of the computer. Quantification is the fourth step and can include counting, measuring, and classifying objects, as well as assessing density, texture, and other features. These results can then be exported to a spreadsheet, such as Lotus or Excel. In the case of MOCHA, the data can be either exported to a spreadsheet or managed within the program by a data worksheet that is compatible with Jandel Scientific's SigmaPlot scientific graphing software. The fifth step is to manipulate the data, for either analysis or presentation. Such manipulation can include calculating statistics and generating charts and graphs. The Macintosh platform took an early lead in PC-based image analysis, largely on the strengths of its GUI, according to researchers. Now, however, the Windows operating system, with its GUI, is changing the terms of engagement for many IBM-compatible PC users. One result has been the appearance of these more accessible image-analysis packages that take advantage of the acknowledged computational power--especially per dollar invested- -of the 386 and 486 microprocessor machines. "With the number-crunching that you get out of a 33MHz or a 50MHz 486, it's almost like real-time image processing," says Chris Sieracki, a graduate student in optics and electrical engineering at Dartmouth College's Thayer School of Engineering, Hanover, Mass. "And PCs are getting more and more user-friendly because Windows is so much like the Macintosh environment. Also, prices have come down, so you can get an awful lot more." Using GLOBAL LAB Image, Sieracki supports a variety of cell studies being conducted by biomedical researchers at the Dartmouth-Hitchcock Medical Center. "Often, if you want a more powerful system," says Christine Wilson, a Data Translation representative, "the Macintosh systems are out of the price range of a lot of laboratories. Now, with Windows, you have power, ease of use, and affordability." "The advantage over the Macintosh platform--and this is not to slam Macintosh, because we use them for other things--is that the price for performance, the bang for the buck, is better on a 486," says Goodwin, an OPTIMAS user. "Also, I love the Macintoshes for doing graphics, but for doing hard-core image analysis, they just aren't fast enough. And that's why we use Windows." Goodwin's image-analysis applications at Fred Hutchinson, an institution known for its development of bone marrow transplantation techniques, include testing potential donor and recipient cells for compatibility with each other to preclude graft-vs.-host disease. "You have to make sure that the bone marrow that you're going to give someone is not going to turn around and kill that person," he says. "We take lymphocytes from the donor and the patient, put them in a dish together, and ask the question: `Do they kill each other?' It's that simple." Two dyes, acridine orange and propidium iodide, which fluoresce green and red, respectively, are used to help answer the question. Acridine orange enters and dyes each cell nucleus, while propidium iodide dyes only the nuclei of the dead cells. A green and a red image are then captured, contrast thresholds are set to enhance the images, and the image-analysis software counts the live and dead cells. "The whole process takes me less than a minute," Goodwin says. "The manual method is, you have some poor guy under a microscope with a bean counter going, `green-green-green-red-red-green-red- green . . .' And after a couple of hours, he says, `Eureka! I have an answer--and I'm resigning tomorrow, by the way, because I can't do this any longer.' " Michael Tainsky is a molecular and cell biologist at the M.D. Anderson Cancer Center in Houston. He uses MOCHA to screen differentially expressed cDNA in cells, applying a technique similar to Goodwin's. "You may have anywhere from 2,000 to 10,000 spots on a plate," Tainsky says. "And the computer just goes, `bink.' So, you can see the difference in efficiency."

Image Analysis Under Windows: New Tools For Biologists

Interested in reading more?

Become a Member of