Chemical cameras reveal a world that is invisible to the human eye. Smaller and cheaper devices are now finding uses from agriculture to cancer diagnostics.
When we admire the vivid array of colours in a rainbow or sunset, how can we not believe that humans have amazing eyesight? Still, there is an awful lot we miss.
Our eyes see colours by gathering light in three broad ranges of wavelengths. Yet every material absorbs and reflects light in its own way, often beyond these ranges, producing a richness that goes unnoticed by the human eye. Now scientists can use advanced cameras to view this wealth of information, revealing not only a material’s colour but also its chemical and biological composition.
“Different materials reflect light in different ways,” says Steve Marshall, director of the Hyperspectral Imaging Centre at the University of Strathclyde (UK). “If you were looking at different pharmaceutical tablets, they might both look white to the eye but they would have a different spectral profile.”
If a picture is worth a thousand words, each hyperspectral image is a novel. Industry and academic researchers can now use special “hyperspectral cameras” to analyse this full spectrum of light and produce images that fingerprint the chemical and biological nature of their subjects. This is useful for a wide range of applications – including sorting plastics for recycling, detecting gas leaks and even diagnosing cancer.
Eye in the sky
Hyperspectral cameras were first developed in the 1980s by military and government agencies like NASA, which used the cameras for remote sensing. Mounted on light aircraft or satellites, detectors mapped the Earth’s surface not just in the visible spectrum, but also in the infrared spectrum to monitor different types of vegetation and minerals from the sky.
“What’s made a difference now is that cameras are smaller in size and cost, which means they can be used in manufacturing and labs,” says Antonio Plaza, who leads the Hyperspectral Computing Laboratory at the University of Extremadura in Spain. In the past the speed at which satellites could function was hampered by having to send huge amounts of data back to Earth for processing. Plaza’s team works on software that can do the job in real time. Rather than examine each band of an image, the software searches the spectrum for patterns that match the fingerprints of different materials. “If you have a satellite collecting data over a forest fire and you want that data to be useful, you have to be able to use it quickly,” Plaza explains.
Hyperspectral cameras are now having an impact across a broad variety of industries. One is food and drink, where infrared images can detect the tenderness of meat or the taste of a sponge cake by analysing sugar, fat and moisture content. Another is medicine, where chemical changes can point to skin cancer (see Applications Snapshot image on the right). “It’s the ultimate non destructive testing,” says Marshall.
Advanced remote sensing cameras are also becoming permanent features of the sky, says Uta Heiden of the German Aerospace Centre. When Germany launches its Environmental Mapping and Analysis Program satellite in 2017 it will carry the latest generation of camera, analysing data across 224 bands of the spectrum in 30-metre wide pixels.
The -200°C camera
Most applications of hyperspectral cameras focus on the visible and near-infrared range. Longer wavelengths – so-called mid or thermal infrared – also contain useful information, though detecting them has been difficult. The tiny vibrations of any object at room temperature give off radiation in this part of the spectrum, confusing sensors with spurious signals. To suppress this noise, cameras usually had to be cooled to -200°C, making them bulky and expensive.
Breakthroughs in technology have reduced the cost of imaging in this region, allowing new uses. While conducting experiments with laser light in 2007, Christian Pedersen and his colleagues at the Technical University of Denmark chanced upon a way to make low-energy infrared waves detectable in the visible region. They saw that by crossing a laser beam with the light from the target inside a crystal they could boost the combined signal into the visible range, while maintaining the image. “The laser gives all these mid-infrared range photons a kick, shifting them and their imaging information so that they can be registered by an ordinary camera,” explains Pedersen, who co-founded the IRSee company in 2012. The company’s detector can pick up a single infrared photon, showing structural changes in tissue that can be used in cancer diagnostics; it can also reveal otherwise invisible gases, such as methane and carbon dioxide.
These potentially noxious gases are also the focus of Rebellion Photonics, a Texas-based start-up. Traditional hyperspectral cameras analysed an image line by line, with each shot taking a few seconds to generate (see The Camera as Detective). This prevented them from capturing moving objects, according to Robert Kester, who co-founded the company as a graduate student at Rice University in 2009.
Kester’s brainwave was to use a group of tiny mirrors to redirect light from each pixel onto an array of “sub-imagers”, which reduce each pixel’s light down to a slim line. Each of these lines is then sent though a prism, where it spreads into its composite wavelengths, so that all the information can be captured on a two-dimensional detector. The technique allows the camera to produce an instant snapshot of the entire image. The camera captures 30 images per second – so fast that it in effect becomes a hyperspectral video camera. The device then overlays false colour images of any gas clouds it sees over a regular video stream. This allows oil and gas companies to detect leaks as they occur, replacing safety technology that dates back to the 1950s. Last year the company won The Wall Street Journal’s Start-up of the Year competition – with the paper praising its goal of revolutionising the $10 billion rig and refinery safety market.
Hypervision for all
Cameras are not only becoming quicker and more sensitive, but also ever smaller. Visnx, a company that grew out of the École Polytechnique Fédérale de Lausanne (EPFL) in Switzerland, promises to reduce the weight of a hyperspectral camera from kilos to a few grams. “Our technology allows us to minimise the sensors so that we can put them on almost anything,” says EPFL’s Yosef Akhtman. That includes unmanned aircraft – useful for Akhtman’s project creating hyperspectral maps of lakes in Switzerland and Siberia. They can also attach to standard microscopes and become portable.
Timo Hyvärinen of the Finnish company Specim, which manufactures hyperspectral cameras, predicts that within two to three years handheld versions will allow everyday professionals – farmers, doctors, police officers and environmental inspectors – to instantly access this invisible world.
Seeing this chemical world in everyday life could open up a host of new opportunities, says Pedersen. “It gives you a whole new set of eyes.”
Colours emerge because materials absorb light differently, leaving some wavelengths to reflect more into our eyes than others. Human eyes are sensitive only to broad overlapping ranges of frequencies that peak around red, blue and green, three colours that cover all the wavelengths in the visible spectrum.
Hyperspectral cameras, on the other hand, can detect many different wavelengths separately. They can also see across a wider spectrum than humans can, extending into infrared and ultraviolet.
When light enters the camera, a prism splits it into its constituent wavelengths. Detectors measure the light in each of these hundreds of narrow bands to see how the material reflects and absorbs light over the full spectrum. This can be used to learn about the material: what kind of atoms it is made of and how they are bonded.
Traditional hyperspectral cameras scan an image line-by-line, like a printer churning out text. As each pixel comes with a tower of data, every line comes with its slice of a “data cube”. Once the full image is captured, each horizontal cross- section shows the entire image in a single wavelength. Innovations in “snapshot” imaging now let some cameras capture both the spectrum and position of each pixel at once, allowing hyperspectral cameras to take many images per second and making hyperspectral movies.
When it comes to colour vision, not everyone is born equal. Recent studies show that a small number of women carry a genetic abnormality that gives them an extra primary colour with which to mix their palette.
Give these women two colours that most of the world would describe as identical and they will see them as different. Yet explaining what they see is difficult, says Newcastle University neuroscientist Gabriele Jordan. “They might say there’s more gold in one than the other, but it’s arbitrary. There just aren’t words for the colours normal people can’t see,” she says.
Such people are called “tetrachromats”, because their colour vision is derived from the brain interpreting signals from four instead of three receptors. This usually results from genetic abnormalities that shift the peak sensitivity of the red and green receptors.
Only women can be tetrachromats because the abnormal genes sit on the X- chromosome. A man inherits one X-chromosome, while a woman gets two. When the gene for one of these cones mutates in a man (which happens to some degree in 8 per cent of men), the red and green cones come closer than normally, making it harder to distinguish between wavelengths in this region. This is why men are more prone than women to colour blindness. When a woman inherits such an abnormality, she has a second X-chromosome to fall back on.
Animals already see a wider range of colours. Although most mammals have only two cones, birds usually have four. Many insects can see ultraviolet light; some flowers have even developed patterns in these wavelengths that attract honeybees – though not humans.
By Elizabeth Gibney (reporter for Nature in London)