Pfeiffer’s approach could be used to improve security systems too. Conventional x-ray imagers like those at airport-security checkpoints can’t differentiate between many different kinds of materials–for example, chocolate and cheese appear identical to some explosives. But cheese and explosives scatter x-rays differently, so in Pfeiffer’s dark-field images, the differences between the two materials are apparent.
Pfeiffer has already begun making CT scans with conventional x-ray tubes using another contrast-enhancing technique he developed two years ago, called phase contrast. He says that he’s currently working to incorporate gratings for dark-field imaging into conventional CT devices. He’s also collaborating with researchers at the Center for Biomedical Imaging, an institute run by the University of Lausanne and the University of Geneva, to determine whether dark-field x-ray imaging can be used to tell healthy tissue from cancerous tissue. Cancers don’t absorb x-rays very differently than healthy tissue does, so x-ray systems that rely on other properties, such as scattering, might make for better mammograms, for example. Lanza’s group at MIT is also working to develop better cancer-detecting CT scanners that use a combination of absorption and refraction for contrast and also rely on nanofabricated gratings. (See “Changing the Physics behind X-Ray Imaging.”)
Dark-field imaging has been used for more than 20 years to enhance contrast and resolution in conventional optical microscopes. But applying the contrast-enhancing techniques that work well with visible light to x-rays has taken a long time, says Pfeiffer. Such a system is only now possible thanks to advances in photolithography and many years of basic science research using synchrotrons, he says.
Pfeiffer envisions that future x-ray imaging systems will be like what light microscopes are today: they will incorporate many complementary systems for enhancing contrast–absorption, refraction, scattering–and doctors and researchers will be able to use whichever combination works best for a given sample.