Applications Of Image Analysis Systems Expand Beyond The Research Lab
TIME EFFICIENT: The AMBIS radioisotopic imager from Scanalytics/CSPI. Already an invaluable tool in some basic research, image analysis is edging into the classroom and the clinic. "Any field of life science that can put a video camera onto a microscope will begin to use image analysis," predicts Richard Cardullo, an associate professor of biology at the University of California, Riverside. In general, the technique acquires, digitizes, and then processes a microscope or scanned image, enhan
Oct 28, 1996
Ricki Lewis
 TIME EFFICIENT: The AMBIS radioisotopic imager from Scanalytics/CSPI. |
In general, the technique acquires, digitizes, and then processes a microscope or scanned image, enhancing some areas and subduing others, so that the user can measure, quantify, and otherwise analyze the data. Image-analysis systems can also archive data, which is particularly important in clinical trials.
 COMPLETE PACKAGE: The KS 300 system from Carl Zeiss. |
Researchers and suppliers report that several advances are bringing image analysis to more laboratories. "Computers are getting faster, cameras are getting more sensitive and quantitative, and image-processing hardware and software are getting better and cheaper by the day," Cardullo says.
Kohlhaas mentions a device called a frame grabber to illustrate plummeting prices. A frame grabber is a printed circuit board installed inside a computer to convert analog signals, such as from a video camera, into discrete digital information. "Several years ago, a frame grabber cost $20,000 to $30,000. Two to three years ago it was $4,000 to $8,000, and now it is $600 to $800," Kohlhaas says. Why Analyze Images?
Cardullo, like many researchers, uses image analysis to track events that are so fast and complex that their details can elude the human eye. He investigates fertilization in mice. "The cells we work with exist in a number of different physiological states at any given point in time. This makes routine biochemical and biophysical measurements very difficult to take and data difficult to interpret," he explains. "The overall goal of image analysis in our laboratory is to make biochemical and biophysical measurements at the level of the single cell." Cardullo uses a combination of commercially available and homemade equipment for image analysis, based on a Zeiss Axiovert microscope.
For Rhona Schreck, director of cytogenetics at Cedars-Sinai Medical Center in Los Angeles, the advantages of image analysis are numerous. Cytogenetics laboratories typically crank out karyotypes, which are charts that align chromosomes in size-ordered pairs to reveal abnormalities, usually in fetuses. In times past this was done manually, by collecting, staining, photographing, and developing chromosome images, then cutting and pasting them into a chart. Automatic karyotyping replaces that approach.
"The CytoVision Karyotyper system eliminates photography and darkroom work, facilitates data storage and retrieval, and speeds the process of producing a karyotype," Schreck says, referring to a system from Applied Imaging Corp. of Pittsburgh. Automated karyotyping is faster, gives higher resolution, and permits image enhancement. "These systems are invaluable for fluorescence in situ hybridization [FISH], allowing the visualization and documentation of faint signals that could not be recorded by traditional photography," she adds, cautioning that it still takes a highly trained eye to spot abnormalities, even in an enhanced image.
Fans of image analysis like to quantify the savings of time and space that the technology affords them. For a clinical trial of a lymphoma treatment, for example, Bio-Imaging Technologies Inc. of West Trenton, N.J., is transforming standard CT scans of tumor shrinkage over time into computerized three-dimensional reconstructions. "Otherwise, we would have to deal with 12,000 films on light boxes!" remarks Peter Langecker, vice president for clinical research at Palo Alto, Calif.-based Coulter Pharmaceuticals Inc., the company submitting trial data to the Food and Drug Administration (FDA).
Similarly, archiving eliminates such mundane tasks as "going to the cabinet and spending a day digging through hundreds of images," explains Kohlhaas. "A medical researcher can store images in different folders and add text, such as a diagnosis. Later, to retrieve particular images, such as muscle fibers, the user simply types 'muscle fiber.' All the images with that key word pop up."
An image-analysis system consists of a host computer and an image-processor board interfaced with a camera or scanner. (A camera forms a simultaneous multipixel image, whereas a scanner forms an image one pixel at a time.) A camera typically captures an image from a microscope field, and a scanner might depict electrophoresis gels or medical scans. The image processor translates analog information, such as color and contrast, into digital shades of gray that the computer can store. Software programs then analyze the data. Most programs run on Windows 95, but some companies, such as Signal Analytics Corp. of Vienna, Va., service Macintosh systems.
 VARIED APPLICATIONS: The genomyxSC Scanner |
 ECLECTIC: The chemImager system from Alpha Innotech. |
 FROM LEICA: The Quantimet 600 is also a complete system. |
Some complete systems are targeted to specific applications. This is the case for Downer's Grove, Ill.-based Vysis Corp.'s Genetics Workstations. The Quips LS Genetics Workstation handles automated karyotyping and FISH and costs $59,900. The Quips XL Genetics Workstation sells for $87,900. In addition to doing what the LS workstation does, the XL model conducts comparative genomic hybridization (CGH), a method to detect genetic abnormalities.
 IMAGEMASTER: Photon Technology International's system for quantitative video fluorescent imaging. |
Quantitative ratio fluorescence imaging requires two wavelengths of incoming light, and a detector that can pick up two returning wavelengths. "The heart of the system is the illuminator, which provides the alternating excitation wavelengths necessary for ratio imaging," says marketing manager David K. Smith. In a typical experiment, a cell is stimulated, and a series of images are acquired, stored, and analyzed. The price of the system ranges from $30,000 to $50,000, Smith adds.
The InCa++ System, one of several from Intracellular Imaging Inc. of Cincinnati, is also a complete fluorescence imaging system. It includes a fluorescence microscope, a low-level-light charge-coupled device (CCD) camera, a UV/visible light source, an automated filter changer, and a Pentium computer with all the necessary software (R. Finn, The Scientist, Feb. 19, 1996, page 17). Prices for the InCa systems, which all come complete, start at less than $18,000 for single-wavelength and less than $30,000 for radiometric systems.
Fluorescence ratio imaging systems for Macintosh systems are also available from Signal Analytics Corp. In addition, the company designed a Macintosh-based image-analysis system, called the Image Explorer, specifically for Okemos, Mich.-based Meridian Instruments Inc. The Image Explorer system includes a video CCD camera, grayscale frame grabber, and IPLab Spectrum image analysis software for use in several applications, including chemiluminescence, FISH, and microscopy.
Another product from Meridian Instruments is the INSIGHT Point Laser Scanning Confocal Microscope. The system allows confocal images to be viewed through oculars in real time. Its INSIGHT-IQ software provides fluorescence quantitation and image analysis, making it possible to create 3-D reconstructions, including simulated fluorescence process and image animations.
The company recently introduced the Meridian TR Series, which is the first laser scanning confocal imaging system designed with an inverted microscope and single mirror optical path, according to Meridian. The systems, which are designed for Pentium-based PCs running Windows 95, allow simultaneous acquisition and display of up to four reflected light channels and one transmission channel. Image processing, 3-D scanning and reconstruction, and kinetics scanning and analysis packages are also available.
Foster City, Calif.-based Axon Instruments Inc. brings image analysis to neuroscience with products that allow researchers to combine imaging with electrophysiological measurements of the ion fluxes that underlie neural activity. The Axon Imaging Workbench is a software program that coordinates instruments and devices that acquire images of ion conduction. It costs $3,000. The Digidata 2000 Image Lightning, also from Axon, is an image-processing board that sells for $4,200.
Axon Instruments' products give neuroscientists new capabilities. "We can now monitor the changes in intracellular ions and analytes with the use of fluorescent indicator dyes. We can measure activity along cellular elements that are too small to be accessed by standard electrophysiological techniques, and we can measure activity among populations of cells simultaneously, with high spatial and temporal resolutions," explains David Wellis, an applications scientist at the company. "None of these are possible with standard electrophysiological experiments."
Another popular Axon Instruments software product is PCLAMP. Jon Lederer, a professor of physiology at the University of Maryland Medical School in Baltimore, uses PCLAMP to acquire and measure voltage clamp data on fleeting calcium ion fluxes called calcium sparks. "Subcellular increases in calcium ion concentration occur very quickly in a very small space of about 2 microns diameter," he notes. "The time to the peak is about 10 milliseconds. This can't be seen by conventional calcium imaging."
 CAPTURES IMAGES: Neurolucida computer imaging system from MicroBrightField Inc. |
To the untrained eye, an electrophoresis gel may seem little more than a lane of smudged bands. Image analysis can objectively correlate band positions to DNA or protein fragment sizes, and determine optical densities to calculate concentrations. Furthermore, digital archiving enables researchers to easily send gels to each other and to prepare gels for publication. The digital route is far less cumbersome and time-consuming than traditional photo documentation of research results.
Image-analysis systems are available for both fluorescent and radioisotope-based electrophoresis experiments. The FluorImager SI, a product manufactured jointly by Molecular Dynamics Inc. of Sunnyvale, Calif., and Amersham Life Science Inc. of Arlington Heights, Ill., provides quantitative fluorescent imaging of gels, blots, thin-layer chromatography, and microtiter plates. The heart of the unit is an optical system (scanning module) that passes a laser over a fluorescently labeled sample and measures emitted light, generating an image of the sample. FluorImager SI also includes software and reagents, and works on a Pentium-based computer or a 486DX.
Scanalytics' AMBIS radioisotopic imaging systems detect radiolabels emitted from gels, blots, and plates. This takes minutes, compared with the conventional method of indirectly detecting radioactivity by exposing X-ray film, which takes days. Scanalytics is a division of Billerica, Mass.-based CSPI Inc.
Image-analysis systems can augment or sometimes replace other technologies, such as flow cytometry or confocal microscopy.
The HTM-ALTOS optical scanning cytometer offered by Hamilton Thorne Research of Beverly, Mass., rapidly identifies specific cell types within up to 1 million cells, according to the company. "The HTM-ALTOS fits in a niche between flow cytometry and image analysis. Flow cytometry doesn't really like as few as 500,000 cells, and it isn't accurate if you are looking for 10 to 20 cells," says Meg Douglas Hamilton, CEO of Hamilton Thorne Research. The cytometer is useful in FISH, for identifying chromosomes that bind DNA probes, and for diagnostic tests for cytomegalovirus.
HTM-ALTOS is a proprietary internal optical system, with magnification as good as or better than that of a microscope, Hamilton claims. "It automatically goes from one slide to the next, scans the whole slide or just part, and then produces an edit page with all selected cells. A user can scan a field and highlight all tagged cells," she adds. Prices range from $30,000 to $45,000, depending on software.
Software from VayTek Inc. of Fairfield, Iowa, can substitute for or improve confocal microscope images. A confocal microscope removes the haze that comes from out-of-focus planes by passing light through small apertures. VayTek's HazeBuster and MicroTome software packages take images from a standard light microscope that have been captured by a video camera and digitized, then apply a deconvolution algorithm that cleans up the image.
The primary advantage of VayTek's software is cost, compared with the $75,000 to $300,000 price tag of a confocal microscope. HazeBuster sells at $7,500 for Windows and $8,500 for Macintosh. The faster MicroTome costs $16,950 for Windows and $14,495 for Macintosh. Another VayTek product, VoxBlast, adds volume to images and sells at $2,995 for Windows and Macintosh.
| Alpha Innotech Corporation Amersham Life Science Inc. Applied Imaging Corp. Axon Instruments Inc. Bio-Imaging Technologies Inc. Genomyx Corp. Hamilton Thorne Research Intracellular Imaging Inc. Leica Inc. Meridian Instruments Inc. MicroBrightField Inc. Molecular Dynamics Inc. Photon Technology International Inc. Scanalytics/CSPI Inc. Signal Analytics Corporation VayTek Inc. Vysis Corp. Carl Zeiss Inc. |
In the biology department at Chatham College in Pittsburgh, visiting instructor Michael Zumpano uses VoxBlast to analyze CT scans of fetal skulls to study craniofacial development. "VoxBlast allows you to manipulate 3-D volume while viewing individual 2-D CT slices," he reports. "Those of us interested in modeling growth need to be able to locate three-dimensional landmark data for points on the skull, and to measure angles and the distances between two points. VoxBlast allows this image quantification."
Image analysis is already being used to scan Pap smears, increasing accuracy in detecting cancer cells. Several research groups are developing software to similarly analyze mammograms. Andrew Francis Laine, a professor of computer and information sciences and engineering at the University of Florida in Gainesville, is developing one such system. "Mammograms are acquired through a digital detector and stored in a computer. We then take a digital image matrix and decompose it into more fundamental components," he says. "Essentially we rip the image apart and glue it back together again. The hope is that by breaking the image down, finer details emerge."
Image analysis is entering a new niche in the drug-approval process. In March 1996, Kathryn C. Zoon, director of FDA's Center for Biologics Evaluation and Research, published the Computer Assisted Product License Application Guidance Manual, which instructs drug developers on how to submit computerized image analysis clinical trial data. Bio-Imaging Technologies helps companies seeking FDA approval digitize medical scan data and analyze the images in three dimensions.
Image analysis not only speeds evaluation of results, but also can reveal more information than can traditional clinical endpoints such as assessment of patient survival, or two-dimensional analyses of a tumor's greatest diameter. "What's changed is [FDA's consideration of] rigorous assessment of the volume of a tumor after intervention, when compared to the tumor at baseline before treatment. We find that this correlates well to survival," says James Conklin, founder and chief scientific officer of Bio-Imaging Technologies.
And image analysis has value beyond tracking tumor shrinkage. "It can apply to the volume of strokes, and to the sizes of multiple sclerosis plaques. Any disease process that you can observe through imaging technology can be investigated with image-processing tools," Conklin adds.
Finally, image analysis may revolutionize the learning laboratory, according to Robert Blystone, a professor of biology at Trinity University in San Antonio, Texas, who teaches microscopic anatomy. In 1991, under a National Science Foundation Instrumentation for Laboratory Instruction grant, Blystone assembled a digital-imaging workstation for his students, for less than $10,000.
"I purchased two Macintosh 2ci computers, I dropped a frame grabber in, put a digital camera on top of a Nikon or Zeiss microscope, and a student could frame-grab anything he or she could see under the microscope to the computer," he says. Blystone then networked several stations to each other, to the campus computer, and to his students' computers. "They can capture, store, and call up images in their dorm rooms. I can assign microscope homework!" he declares.
Blystone lists what image analysis allows his students to do-measure the angles of bronchi in cats to reveal the mechanical principles underlying anatomy and physiology; manipulate 3-D reconstructions of organs; even morph a chick through development. "What makes this different is that traditional microscopy provides a primarily 2-D image that is static in time. I am able to work in three or four dimensions, and it is interactive and more quantitative," he says.
With prices down and ease of use and choice of products up, it's clear that image analysis will become an increasingly important tool in many life science and biomedical laboratories. Blystone, in summarizing the value of image analysis in his teaching, succinctly pinpoints the technique's value in diverse fields: "Suddenly, computer-assisted image analysis reveals something more."
Ricki Lewis, a freelance science writer based in Scotia, N.Y., is the author of several biology textbooks. She is online at 76715.3517@compuserve.com.
