Matrox Imaging - News & Events

In Living Color

Microscopy tool uses MIL to automate color-based cell analysis

By Sarah Sookman, Matrox Imaging

In order to determine the most appropriate treatment for diseases such as cancer, researchers must study individual cells and tissues to determine a therapy's effectiveness. Their analysis relies on hundreds of immunohistochemical (IHC, tissue-based) and immunocytochemical (ICC, cell-based) tests, which involve staining samples with commercially available reagents, and observing the sample's chemical reaction under a microscope. With the seemingly infinite number of cells that must be studied by a pathologist, it is easy to appreciate the time, effort, and patience required to test some 400 new cancer therapies that are currently in development.

Image processing techniques have been used in microscopy for some time, because manual microscopy leaves room for error. In 2000, Chromavision (San Juan Capistrano, California) put its Automated Cellular Imaging System (ACIS®) on the market. The ACIS® automates tasks for the pathologist, improving the accuracy, precision and reproducibility of slide-based testing. ACIS combines an automated microscope and digital camera with commercial and proprietary hardware and software. "The system detects, counts, and classifies cells based on color, size, and shape and provides quantitative information and standardized testing results that are impractical or even impossible to deduce with the human eye and a microscope alone," explains Judy Kropp, Marketing & Communications Manager. The pathologists perform tests to determine if disease is present, and then use those results to recommend the appropriate treatments to the attending physician.

Some of the technology used in the ACIS system is derived from technology used in the Strategic Defense Initiative (Star Wars). Part of the Star Wars program used color analysis to detect live missiles from decoys as the missiles entered the atmosphere. Since almost all laboratory tests rely on some type of color analysis, the device was further developed for slide-based testing. Today, over 200 ACIS systems are currently in operation in the US; additional systems are in use across Europe.

Many of Chromavision's clients use ACIS for breast cancer testing, although the system is capable of automating an unlimited number of tests in clinical and research environments. ACIS performs four main procedures, all of which aid the pathologist in choosing the best treatment for the patient. (See sidebar.)

The system

In the system, a Matrox Meteor-II /Multi-Channel frame grabber is connected to a progressive scan, 60 fps three CCD color camera that is mounted over the microscope's optics. Typically, the camera grabs several thousand fields of view for each slide, beginning with the microscope's lowest power. Image processing algorithms analyze each field of view, and if the processing detects something of interest such as a diseased cell, the microscope resets its optics to a higher power, and the camera will grab more images for further analysis. If the analysis suggests some manifestation of disease, referred to as a positive event, the field of view will be displayed. When the frame grabber has acquired images at the microscope's highest power and analysis is complete, the pathologist reviews the displayed fields of view to make a diagnosis or prognosis.

Much of the image analysis is performed by the Matrox Imaging Library development toolkit, using algorithms for image processing, blob analysis, pattern matching, and measurement. Proprietary software is responsible for managing the cell images, such as locating a target or a flagged cell. In staining measurement analysis, other proprietary algorithms calculate the intensity and convert it to a certain number of proteins. All the images are displayed in a large GUI for the pathologist. "We call it a montage, a group of single cells that we display," says Jeff Caron, Senior Software Engineer. "You have 60 cells on one page. Then you can use your mouse or keyboard to mark, accept, and reject cells from that flagged group so they are included or excluded from the final lab report."

If a slide renders a thousand images in a small field of view, the pathologist might have to look at those cells in a larger region. Chromavision has a patent on its histological reconstruction algorithms, which reassemble all the fields of view from the slide. The reconstruction functionality is especially useful for intensity-based analysis. "You can look at the tissue and using our tools, highlight an area, and ACIS will give you an intensity score which correlates to an amount of protein in the cell," explains Caron. The pathologist can then use the reconstruction to see how the cells are distributed, or investigate how the tumor is invading an area of the sample. "The reconstruction allows the pathologist to see the slide's sample in a larger frame of reference than what is possible with a microscope, " adds Jean Smith, Manager of Software Engineering. "The pathologist can choose areas of interest in that tissue, areas where the tumor is expressed more sufficiently."

Up to 100 slides can be placed in a carrier at one time, and a custom-designed robotic mechanism loads the slides on and off the microscope's stage. Running completely unattended, the ACIS scans those slides, processes the images, and displays all positive results. Then the carrier can be loaded with the next batch of slides.

When slides are examined with the human eye and a microscope, missing a small number of cells is a likely possibility. ACIS acts as a pathologist's assistant; its eyes never tire, and it brings positive results to the pathologist's attention. From the thousands of images rendered per slide, ACIS might display only 60 for the pathologist to manually inspect. When the analysis is complete, the pathologist prints a report which includes full color images of the cells and tissues along with numerical results based on the algorithms. The report is flexible enough that the pathologist can enter individual comments, making it a valuable communication tool for pathologists, oncologists, and their patients.

Challenges in development

While ACIS is a versatile tool, obtaining good results depends on the quality of the slides. Analysis requires images that can be well segmented. Obtaining good results depends on color threshold, as well as having good samples from the lab. Indeed, one of the challenges during ACIS's development was inconsistency in laboratory procedures. "With every lab doing slide preparation a bit differently," recalls Caron, "sometimes the cutting or preparation of the tissue can cause the stain to be absorbed a little too much, or not enough." In a system where color intensity is vital for extracting results, these procedural differences undoubtedly cause problems. "There has to be a synchronicity between the lab and what the ACIS can do," notes Smith.

Chromavision resolves potential inconsistencies by reviewing slides from clients. "We look at their slides to determine how they can modify their staining procedures to optimize them for image analysis, if need be," says Kropp. The technical services team works to make sure that the staining is the best possible and the most standardized. "What our instrument adds is the quantification that can't be done with the human eye, and standardization of scoring and reporting of results," she adds.

The ACIS system offers increased accuracy, precision and reproducibility for pathological examinations, and improves communication between pathologists and physicians, but Chromavision is more than a software company hat produces tools for image analysis. "We are a full-service company," says Smith. "We have the technical services team, and a fully-staffed lab with pathologists and lab technicians under the same roof as hardware and software specialists. ACIS becomes a real system that way, and that is what makes us different from other companies doing this kind of work."

For more information, contact our Media Relations Team.