Bioscience imaging faces three challenges - ease of use, diversity, and meaningful quantification.By Mike May
A needle slips inside an animal, guided to a small spot-a blur really-seen on a computer tomography scan of the animal's lung. The needle contains a tiny microscope. When it reaches the "blur," the scope scans the individual cells, stained with a dye, aiming to determine if the spot is cancer or not. If it is, the same needle can send out radio frequencies or other ablation techniques that destroy the cancerous cells. While this is not yet possible in humans, it could be, and it is an ongoing project in the laboratory of Stephen Wong, PhD, PE, JS Dunn distinguished chair professor of biomedical engineering, chief of medical physics, and vice chair of the department of radiology at The Methodist Hospital, Weill Cornell Medical College, and the director of the bioinformatics program at The Methodist Hospital Research Institute in Houston, Texas. "This would be minimally invasive medicine," says Wong. "With this, maybe we can detect and eliminate cancerous cells before they turn into a tumor." Wong is already testing this system on animals.
 Taken from automated time-lapse imaging for high-content screening, this image shows breast-cancer cells from an HER2-positive SKBR3 cell line. The cells have been treated with compounds and stained (blue: nuclei; red: microtubulin; green: actin). The image was acquired with an automated fluorescence microscope custom-built by Olympus. (Source: The Methodist Hospital) |
In bioscience imaging and analysis, Lon Nelson, marketing manager of Leica Microsystems' life science division in Chicago, sees three main challenges. The first is ease of use. "We are trying to make microscopy available to the masses," he says. Next is diversity. "A core imaging facility sees samples that are more widely varied now than ever, and it will get even more diverse," Nelson says. He points out fixed and live cells, different fluorescent proteins, and so on. The third challenge is meaningful quantification. "In biology, meaningful measurement can mean different things to different researchers, but certainly the trend is towards increased quantification," he says.
As researchers take on those challenges, a variety of obstacles must be negotiated.
Areas of Improvement
For ease of use, Nelson says that Leica Microsystems uses software to help guide people through experiments. The company is also reducing the complexity of techniques, such as stimulated emission depletion (STED) microscopy. As Nelson says of STED, "It started on an optical bench with lasers going everywhere." Now the Leica TCS STED comes in an integrated package, and it can dip below 100 nanometers in resolution-into so-called super-resolution.
Other optical companies are also focusing on ease of use. For example, Nicolas George, group manager for light microscopy in the marketing department at Olympus America in Center Valley, Pa., describes the Fluoview FV10i as "the first self-contained, laser-scanning confocal microscope." He adds, "It's very simple to operate." In fact, it doesn't look like a microscope at all-more like a small, streamlined computer printer. There aren't even eyepieces. Instead, a scientist views images solely on a computer screen. "The microscope brings confocal imaging to the masses," George says.
 These breast-cancer cells have been labeled for several proteins, shown here with conventional color, CRi's Nuance multispectral, and CRi's inForm segmented images. (Source: British Columbia Cancer Agency) |
Likewise, Olympus recently introduced an easy-to-use fluorescent scope, its FSX100-another microscope with no eyepieces. "This is for people with little to no microscope experience," George says. "Maybe a scientist has only been looking at cuvettes or gels, whereas the use of a microscope would make conclusions more powerful by adding images."
More to See
To help with diversity, Leica Microsystems developed a tunable white laser for confocal microscopy. "You can use this with any fluorescent protein that excites from 470-670 nanometers," Nelson says, "and you can use up to eight at once."
In confocal microscopy, the scanner also impacts the resulting image. Nikon's A1 series, for example, includes non-resonant and resonant scanners. "With the resonant scanner," says Mike Davis of Nikon Instruments in Melville, N.Y. "You can scan at over 400 frames per second." This scope also includes various detection modes, including one for spectral imaging, which can be used to track multiple probes that have overlapping fluorescent emissions. The A1 series also includes virtual adaptable aperture scanning (VAAS), which Davis says, "collects light that is normally rejected by the emission confocal pinhole." That intensity information can be used later to improve resolution or add back intensity lost due to bleaching during time-lapse imaging.
As scientists track increasingly specific targets-such as a particular fluorescent marker on a certain molecule-the light must also grow more specific, which requires filters. Gregg Fales, product line manager at Edmund Optics in Barrington, N.J., says that a filter's performance depends on three key characteristics: transmission, blocking, and the steepness of the edges. "A high-end filter transmits more than 90 percent of the desired light," says Fales, "and with an optical density of six, only 10-6 of the unwanted light gets through." In addition, a high-performance filter has very steep edges, which are the transitions between transmitting and blocking light. Biologists typically use filters to "dig out a signal from the background," says Paul Gelsinger-Austin, a research and development engineer at Edmund Optics.In January, Edmund Optics released two families of filters targeted at bioscience. One consists of notch filters, which block only light in a specific band. Edmund Optics' new notch filters have an optical density of six. In addition, they are made with Rugate coating technology, which Fales describes as a single-layer coating that has a sinusiodally varying index of refraction. "You get very broad transmission with very deep rejection in the notch," he says. The second new line consists of high-performance long-pass filters, which are available in 50 nanometer steps with an optical density of four.
  Comparing two optical band-pass filters centered at 600 nanometers-one with an optical density of 4 (left) and the other with an optical density of 6 (right)-shows the enhanced transmission of the yellow spheres with the higher-performance filter. (Source: Edmund Optics) |
Better cameras also improve what researchers can see. "One of Hamamatsu's latest offerings in the biomedical field is the ORCA-R2," says Scott M. Blakely, regional product manager for OEM cameras at Hamamatsu in Sewickley, Pa. "Combining the Hamamatsu ER150 CCD, with deep-cooling, very low read noise, dual digitizers (12 and 16-bit), spectral response from UV-IR, and full well capacity up to 36,000 electrons, the ORCA-R2 will produce 1.3 million pixel images at 16 frames per second." He adds that this "camera meets the demands of nearly any imaging application. The ORCA-R2 is extremely flexible in terms of high- and low-light level applications."
Measuring More and Faster
To boost quantification, Leica Microsystems partnered with Molecular Devices (now a part of MDS Analytical Technologies) in Sunnyvale, Calif. This resulted in Leica Microsystems' MM AF software, which runs off of Molecular Devices' MetaMorph. MM AF can provide three-dimensional particle tracking, fluorescence ratioing, and more.Other optical companies also offer advanced imaging software. For example, Carl Zeiss, headquartered in Oberkochen, Germany, developed its AxioVision Physiology software, which grabs and analyzes images. For example, it can use ratiometric calculations from fluorescent images to measure calcium concentrations or pH.In some cases, things work smoother in hardware rather than software. To improve acquisition speed, Nikon turned to direct hardware triggering methods. "A user defines in software a 'menu' of the hardware devices and experiment protocol," Davis explains, "and once the experiment is launched, the orchestration of device control is handled by the hardware devices directly." He adds, "This makes the timing more precise and eliminates 'dead time' communicating back-and-forth with the software during the experiment."
Timing is already a driving factor in microscopy. Scientists always want scopes that grab images faster, especially for live-cell projects where illumination damages the cells. Moreover, says George, "Scientists want more temporal resolution, to see things in real time. By viewing events at time scales of a millisecond or faster, they can better understand dynamic life processes within cells."
Future imaging technology could also enhance high-content screening. At Wong's lab, he is already working on automating this technology-performing time-lapse imaging on 384-well plates with 5,000 cells per well. "Traditional bioimaging people look at a couple pictures," Wong says, "but I can get a terabyte of data in one afternoon." Wong envisions a system in which a researcher or a robot arm inserts a plate and mountains of high-content data emerge. "You need very sophisticated image processing that is not available," he says. For example, he and his colleagues developed a system that can grab 220 quantitative features from a single breast-cancer cell that has been treated with a combination of drugs.
One company, Cambridge Research & Instrumentation (CRi) in Woburn, Mass., found a way to gather more data. Its Nuance camera can grab 10 to 30 wavelengths from an image. "With this camera, you can isolate signals much more effectively and accurately from background signals," says Cliff Hoyt, CRi's vice president and chief technology officer. "Then, you can quantify signals on a per-object or -cell basis." This technology can track four to five proteins with immunofluorescence, according to Hoyt. He adds, "We have customers doing as many as 10, but that takes a really heroic effort." In addition, CRi's inForm analysis software uses machine learning to spot specific events, such as protein expression. Moreover, Hoyt says that this software can pick out tumor cells or look for inflammation-related cells. "It can be trained to identify things at about the perception level of a human," he says.
As we can see, imaging is going faster, gathering more information, and bringing us new things to see every year.