Investigators at the California Institute of Technology have developed two high-resolution, lensless, and fully on-chip microscopes based on optofluidic microscopy.
The systems can give a small group of scientists hundreds or even thousands of microscopes with which to image many cells or small organisms in parallel, and could allow researchers to perform microfluidic imaging of a variety of organisms and cell types, without the need for an unwieldy conventional microscope.
The tools use microfluidic flow to deliver specimens across arrays of micrometer-size apertures defined on a metal-coated complementary metal-oxide-semiconductor sensor, or CMOS, to generate direct projection images.
The first microscope uses gravity-driven microfluidic flow for scanning, and can be used to image elongated objects such as C. elegans. The second system relies on electrokinetic drive to control microfluidic flow, and can be used to image spherical/ellipsoidal objects such as cells.
“We can put tens or hundreds of these microscopes on a chip, giving researchers access to hundreds or even thousands of these microscopes that can be operated in parallel,” Changhuei Yang, an assistant professor of electrical engineering and bioengineering at CalTech, told CBA News this week
For instance, if investigators have many cells that they want to look at both pre- and post-drug treatment, these microscopes can enable them to perform high-throughput parallel imaging of those cells, said Yang.
The systems can also complement currently available microfluidic lab-on-a-chip technologies from companies such as Caliper, Cellectricon, Cellix, Fluxion, and RainDance. Microfluidic lab-on-a-chip technologies are used to detect chemicals, run small chemical reactions, and then look at the results on the chip.
“It is quite remarkable that when we talk about researchers using labs-on-a-chip, there is a need for compact imaging solutions for a lot of the problems that they are working on,” said Yang, who is also corresponding author of a paper describing the microscopes that appears online this week in the Proceedings of the National Academy of Sciences. His group’s OFM technology is meant to meet this need, he said.
Typically, when investigators use microfluidic technology, their need to image their chip leads them to a microscope. “This is sometimes counterproductive because you have this really amazing lab-on-a-chip system, and you need a big, bulky microscope to look at it,” said Yang.
“We think that our device addresses the needs of those who want to look at things in a microfluidic lab-on-a-chip system,” he added. “We offer them a very compact imaging solution.”
For example, investigators at the Georgia Institute of Technology recently announced that they have developed a microfluidic chip that can automate and standardize genetic screens of small, multicellular animals (see CBA News, 6/27/08).
The Georgia Tech investigators said that their chip has five features that can ensure that it operates consistently for an extended period, including its compatibility with any standard microscopy setup.
Yang, who is familiar with the Georgia Tech technology, said his team’s microscopes “could potentially provide [the Georgia Tech researchers] with a very compact, on-chip system, and if you built their system on top of our system, you could do compact imaging at low cost.”
Weapon of Mass Production
Yang said it currently takes about two days for his students to assemble a prototype of the OFM system, and explained that he is currently negotiating with an undisclosed “larger” biotech company to have the tools built on a semiconductor fabrication line. This could allow “hundreds or thousands of these microscopes to be built in a single run, and we could then provide biologists and clinicians with samples of the device, so that they could evaluate it for their respective applications.”
“We think that our device addresses the needs of those who want to look at things in a microfluidic, lab-on-a-chip system. We offer them a very compact imaging solution.”
He added that partnering with a large biotech “is a positive move for this technology because its translation to commercial production is associated with a very high start-up cost … and a big company has the resources and the ability to think over a longer timeline, versus a small start-up.”
The ability to mass-produce the chips would also enable them to be sold for around $10 apiece, said Yang. By comparison, the sensor chips that Yang and his colleagues used cost about $20 per chip off the shelf, and “that is with a huge markup in price, because we are buying it retail rather than wholesale.”
“I am confident that we can come up with something that costs $10 per chip to manufacture, as opposed to something that would cost, for example, hundreds of dollars per chip,” said Yang.
The gravity-driven on-chip OFM system was fabricated on a commercially available 2D CMOS image sensor with a 9.9 µm pixel size. The surface of the sensor was planarized with a 2 µm-thick SU8 photoresist and coated with a 300 nm-thick layer of aluminum.
The CalTech investigators then milled two lines of apertures, each 1 µm in diameter and separated by a single line of sensor pixels, onto the aluminum layer with a focused ion beam machine. The apertures were spaced 9.9 µm apart, and each line consisted of 200 apertures.
To image C. elegans, the investigators uniformly flowed the specimen through the channel and recorded the time-varying light transmission through each aperture as the specimen passed. Each time-scan represents a line profile across the specimen.
The specimen passes the apertures sequentially, so if the speed of the specimen is uniform, there will be a constant time delay between adjacent line scans. The investigators could obtain an accurate projection image of the specimen by shifting the line scans with this delay.
“What we do then is take all the light scans of all the holes, piece them together, and voilá! — we have an image of an object,” said Yang.
“Our motivation for developing our OFM technology was that although the conventional microscope has been around for several hundred years now, the design has not actually changed much,” said Yang.
“We … decided that just trying to miniaturize a conventional microscope is not going to work well, because there are so many optical elements that you need to think about shrinking effectively,” he said. “So instead, we have drawn inspiration from the phenomenon known as ‘floaters,’” which are debris particles that float in the field of view of someone looking at a uniformly illuminated background, such as a clear blue sky.
“The cool thing about floaters is that they stay equally clear, no matter what you do,” said Yang. “This imaging process does not use any lenses. Those floaters are actually very tiny objects, only a couple of millimeters, but when one sees them, one sees them with clarity.”
He said that this suggests that those who want to build a microscope system do not need objectives and lenses, as long as the object they want to image is close to the sensor grid itself.
In the case of floaters, the sensor grid is the retina of the eye. “In our experiment, the sensor grid is the CMOS sensor chip that is very commonly found in a digital camera,” said Yang.
“If we simply take an object and put it on a sensor chip, the image that we get is poorly resolved,” Yang said, explaining that the pixel size on a digital camera is very big, but if researchers are trying to get a well-resolved image, they need much smaller pixels.
“We achieved [better resolution] by coating the chip with a layer of metal, and punching a line of holes on to the layer of metal. The metal blocks light from reaching the sensor. Light can only go through those small holes,” Yang said
The resolution of a conventional microscope can range from 300 nm to 1 µm, depending on the objective of the lens that one uses. “Our system actually gives us 1 µm resolution at the moment, which makes it comparable to a 10x or 20x microscope system,” Yang said.
In 2004, researchers at Harvard used automated microscopy for high-throughput cytological profiling.
In addition to drug discovery and development, OFM can also be used for blood fraction analysis, urinalysis, stem cell screening and sorting, and tumor cell counting.