Search
Menu
Gentec Electro-Optics Inc   - Measure With Gentec Accuracy LB

Getting a good virtual view

Facebook X LinkedIn Email
Hank Hogan

At the National Cancer Institute in Bethesda, Md., pathologist Stephen Hewitt doesn’t spend his days hunched over a microscope — even though he has an excuse to do so. He is chief of the institute’s Tissue Array Research Program lab and in charge of making tissue microarrays for the institute’s researchers and for distribution to the research community.

These arrays are used to discover important targets for cancer treatment, and in supporting this effort, Hewitt has his role. “You’ve got to make images of these little devils, and so I do a lot of imaging,” he said.

Instead of peering through a microscope, though, he uses a virtual microscopy platform. The data can be reviewed immediately or at a later time. It is stored in a searchable digital archive with the ability to move data or to zoom in or out on it as needed.

A tissue microarray consists of paraffin blocks in which as many as 1000 tissue cores are embedded. The cores, which are less than a millimeter in diameter, are extracted from a tissue sample using a hollow needle and then transferred to the block, where they are arranged in a densely packed array. A block is sliced into thin sections, mounted on slides and analyzed. Hundreds of slices can be produced from a single block, allowing a wide variety of tests to be performed.

The program is the largest in the field, and the laboratory specializes in making, teaching about and devising new things to do with tissue microarrays. These two factors combined mean that Hewitt and his colleagues have a great many slides to deal with, each with up to 500 cores.

Thus, he has implemented a number of new technologies. One is virtual microscopy with ScanScope systems from Aperio Technologies Inc. of Vista, Calif. The laboratory has a high-volume T2 scanner that can hold up to 120 slides at a time and a single-slide CS scanner. The T2 is an older, larger and slower version of the company’s XT scanner.

Both the T2 and XT have a 20x objective with 0.75 NA, standard, for capturing the images, and the XT has additional optics that allow doubling of the magnification. Hewitt reported that his T2 takes from two to 20 minutes to scan a slide, depending on the area. It images in 24-bit color and with gigapixel resolution at 0.5 μm per pixel at 20x. The system automatically scans the entire slide stack.

The CS, on the other hand, images up to five manually loaded 1 x 3-in. or two 2 x 3-in. slides at a time. Its standard objective is the same as that of the XT, with the option to go to 403. Hewitt said that he uses this instrument when he needs a digital slide with higher resolution or when he is dealing with a slide that has unusual structures.


This digital slide shows a tissue microarray of small snips from an immunostain of esophageal cancer at reduced magnification. Courtesy of Stephen Hewitt, National Cancer Institute.


The captured data is used in a variety of ways. For example, it can provide a digital archive used by expert pathologists for a standard review. Because the data is filed electronically, there is no need to haul slides out of storage and ship them for examination. During a review, the pathologists can move a digital image just as if they were operating a microscope and looking at a physical slide. The images also are used in analysis, for quantification of biochemistry or to evaluate the effects of a therapeutic agent on a tumor.

AdTech Ceramics - Ceramic Packages 1-24 MR

Hewitt said that one can search for images using an annotation system based on the Web-standard extensible markup language (XML). The annotation contains a link to a particular location, along with whatever text the investigator chooses to embed. Metadata such as its label and acquisition-related information also is associated with each image.

Hewitt recalled that, years ago, the magnification for virtual microscopy systems was a maximum of 20x. Now 40x is routine, providing the level of detail needed by pathologists and clinicians.

Dirk Soenksen, Aperio’s founder and CEO, said that the resolution, and not the magnification per se, was important. “We’re trying to digitize the information on a microscope slide at a sufficient resolution so that a pathologist would find the same level of detail in the digital slide image as on the glass slide,” he explained.

Along those lines, he noted several advances looming on the horizon for whole-slide imaging technology. These enhancements include higher-resolution scanning and the addition of the third dimension, so that users can virtually pan in X and Y while focusing through a digital slide in Z. Another pending improvement is the addition of multispectral imaging. It also is possible that, someday, fluorescence or phase contrast imaging will be added. Such advances will increase the amount of information captured.

Aperio plans to release a product capable of oil-immersion scanning by early next year. It would have 60x magnification and 1.4-NA lenses, and would hence enable much higher resolution than is currently offered. That, along with the addition of the third dimension, spectral information or some other advance, will make files even larger than they are now.

For Hewitt, file size already presents a challenge. His slide images can top 2 GB, making them difficult to move around a network or otherwise manipulate. His experience has been that such large files tend to underscore any already-existing network problem, whether it be a slow server or a bad wire. The solution, he has found, is network optimization and careful attention to bottlenecks.

As for the future, one area Hewitt mentioned is computer-aided diagnostics, a tool that is used in radiology to help clinicians evaluate an image. But pathologists do not have anything similar. The technique will require processing an entire slide, which will be a challenge because of the resulting image size. And, by nature, pathology does not deal with black-and-white data. Instead, things are in color, which Hewitt noted complicates many aspects of the image processing demanded by computer-aided diagnostics. “Color has made it a lot harder,” he said.

Contact: Stephen Hewitt, National Cancer Institute, Bethesda, Md.; e-mail: [email protected], and Dirk Soenksen, Aperio Technologies Inc., Vista, Calif.; e-mail: [email protected].

Published: October 2006
Basic ScienceBiophotonicsMicroscopyResearch & Technology

We use cookies to improve user experience and analyze our website traffic as stated in our Privacy Policy. By using this website, you agree to the use of cookies unless you have disabled them.