X-Ray Computed Tomography Expanding Horizons at Duke
From Anthropology to Aerospace Engineering and Beyond
Question: When is an X-ray machine not an X-ray machine? Answer: When it’s a bridge. And at Duke University (Durham, NC), the school’s X-ray micro computed tomography equipment is arguably among the world’s most unique bridges – one with multiple spans bridging a growing number of very interested disciplines and users.
Housed at the Shared Materials Instrumentation Facility (SMIF) at Duke’s Pratt School of Engineering, the XTH 225 ST micro CT X-ray machine from Nikon Metrology (Brighton, MI) along with Nikon’s 3D reconstruction software was installed in March 2013 and envisioned from the start as a shared university resource. The concept of shared capabilities was the driver for establishing SMIF in 2002, according to Dr. Mark D. Walters, SMIF’s director. “The whole idea is to supply shared resources and equipment among the various Duke departments and research groups as well as the external organizations we partner with,” he says. Said resources include 4,000 square feet of class 100 and class 1000 clean room space, and over 2,600 square feet of specialized laboratory space; including a segregated bio room within the clean room designed for the integration of biomaterials with nano, opto, and electrical devices and structures.
“Researchers at Duke as well as biotech firms, electronic materials companies, government research organizations, and many others have world-class yet cost-effective resources for the characterization and imaging of materials on the micro and nano scale,” he adds.
Cataloging Life’s Diversity
Dr. Doug M. Boyer, assistant professor in Duke’s Department of Evolutionary Anthropology, considered access to micro CT X-ray technology essential when he interviewed with Duke last year. “The research I do relies on micro CT data 100 percent,” he says. “We were really pleased that [Duke’s Trinity College of] Arts and Sciences also saw the acquisition of this equipment as a good investment for the research environment on campus.”
Anthropology, literally the study of humankind, is often perceived as socio-cultural science (as in cultural anthropology and its emphasis on a culture’s beliefs, history, and behaviors). “Then there’s physical anthropology – how diversity in biology of humans and other non-human primates provides evidence for questions about human nature and origins,” Boyer adds. “You probably have a more accurate perspective of the kind of research my colleagues and I do here if you think of it as a subfield of evolutionary biology. Diversity in skeletal and anatomical structure among primates (including humans) is my area of focus. But the approaches I take and the broader implications of the questions I address are directly applicable to biological research generally. Raw data are the measured quantities of anatomical samples, and documenting them is essential for repeatability.”
Micro CT ameliorates a number of difficulties involved with evolutionary anthropology, Boyer further explains. For one thing, there are the skeletal and anatomical samples themselves needing to be cataloged and referenced. Many are one-of-a-kind specimens housed in university and museum collections around the world. Time and travel expenses just getting to them are significant. “If we can post digital images of the bones in our studies, then it takes the field to a new level of accountability: not only can a skeptical researcher re-analyze the measurements I put in my appendix tables, but he/she can directly check the individual measurements I provide. This is impossible (or at least fundamentally impractical) currently.”
As in its well-documented medical experience, X-ray computed tomography not only provides a non-destructive means to image and examine a specimen, it provides details such as porosity and density mapping unobtainable by other means. Thousands of digital images can be produced from a single sample, and each two-dimensional pixel in each image contributes to a three-dimensional voxel as computer algorithms reconstruct 3D volumes. For example, with 3000 images, each voxel in the resulting billion or so is processed 3000 times. The result is a 3-D volumetric map of the object, where each voxel is a 3-D cube with a discrete location (x,y,z) and a density (ρ). Not only is the external surface information known, such as with a 3-D point cloud from laser scanning, but internal surfaces and additional information about what is in between the surfaces from the fourth dimension (density) is provided. Furthermore, “slices” produced by the process and accompanying software can yield much information without destroying the sample. As with the growth of computing power in many applications, what took days a decade ago to assemble 3D micro CT information now takes minutes, yielding much more information to users.
Not only yielding information, but sharing it as well. MorphoSource (www.morphosource.org) is Duke’s initiative to build a digital 3D specimen archive to better enable a worldwide user base to study the diversity of life in its anatomical form. Researchers not only can store, organize, share, and distribute their own 3D data, any registered user can immediately search for and download 3d morphological data sets that have been made accessible through the consent of data authors.
Duke has begun by scanning thousands of samples from its own extensive collections and also those of other institutions including the American Museum of Natural History, the Smithsonian Natural History Museum, and Harvard University’s Museum of Comparative Zoology, among others.
“Digitization of skeletal specimens in 3D (to which you can provide worldwide access) is changing the nature of biological study,” Boyer says. “Retention and sharing of 3D is a problem facing the greater academic community who study one-of-a-kind samples. MorphoSource is taking a data-driven field and applying new means of obtaining and interpreting that data.”
A slightly more lofty goal is to tap the potential for automation of analysis of anatomical structural data on a broad scale. “Right now analysis of molecular data (on DNA, the genetic code) is highly automated. Big data sets are relatively easy to amass because of digital sharing: morphological data hasn’t reach this point, for obvious reasons – scanning is the only way to generate comprehensive numerical representations of bones, but such data have been few and far between until recently. MorphoSource will start to build the large-scale samples needed to bring the study of anatomical structure in-line with the genome,” Boyer says. He is currently working with applied mathematicians and statisticians at Duke to “automate” the measurement and analysis of biological structures. “Another reason why the skeleton is under-studied is that most researchers don’t have the expertise to identify or define relevant measurements. With automated algorithmic routines, we hope to avail morphological data to any interested researcher.”
Training and Certification
Duke not only provides micro CT scanning for the school’s medical, sciences, and engineering departments, it also trains and certifies users on how to use the equipment. “This isn’t a 9 to 5 operation, it’s 24-7,” says R&D engineer and CT specialist Jimmy Thostenson. Users interested in certification are trained in lab safety and procedures as well as equipment operation, working one-on-one with SMIF’s micro CT staff.
Users are not limited to Duke students and faculty; interest is rapidly growing from many local and regional companies and organizations. Industrial users include a biomedical company researching pharmaceutical delivery devices, another inspecting components for microwave radios, and another involved in providing next-generation refrigerators.
The pool of people uneducated to the advantages of X-ray CT is orders-of-magnitude bigger than the educated, one expert has said. And even among the educated, a significant percentage are operating blind, unable to see or make use of the full potential in front of them because of a lack of training or a thorough understanding of the technology. By not only installing and using micro CT, but being actively involved in expanding outreach and education, Duke is at the forefront of changing all that.
Sidebar – X-ray CT Basics
X-rays are at the short end of the electromagnetic spectrum with an average wavelength between 10-8 and 10-12 meters, around the size of water molecules, compared to radio waves with wavelengths that could span a soccer field. There are no radioactive sources in X-ray micro CT; rather electrons are produced from a hot filament similar to a light bulb and accelerated at high voltage, creating a beam of electrons reaching speeds up to 80 percent of the speed of light. The electron beam is focused by a magnetic lens onto a metal target, producing a spot typically between 1-5 µm in diameter. The sudden deceleration of the charged electrons when they hit the metal target produces 99.03 percent heat and 0.7 percent X-rays.
These X-rays emanate from the region where the electron beam hits the target. The size of this spot is referred to as the X-ray spot size or focal spot. In general, the higher the voltage applied, the more power is in the beam, and consequently more power is transferred to the target. The more power on the target, the larger the X-ray spot size, and the more X-ray flux produced.
X-rays travel in a straight line, through the object being inspected, and onto a detector. The object will attenuate some of the X-rays (denser objects attenuating more), leaving only a portion to reach the detector.At low X-ray energies (<60kV), differences of attenuation along the X-ray path to the detector are detected and shown as a shadow image. At higher X-ray energies (60-225kV), attenuation and scatter occurs. This scatter generally reduces contrast in the image. With X-ray energies above 225kV, scatter becomes an increasing problem for certain detectors. Above 225kV, scatter can be rejected from the detected signal by a linear detector, although throughput decreases (fewer images per hour). And at greater than 300 - 400kV, scatter is the dominant contrast mechanism, i.e., more X-rays leave the beam from scatter than from absorption.
Amorphous silicon flat-panel detectors have a fluorescent screen which converts the X-ray energy into light to form an image on an array of light-sensitive diodes. Electronics allow this image to be read by a computer. These panels can have pixel sizes over a wide range and sensitivities up to 16 bits (64k grey levels).The resolution of the detector relates in part to the size of the X-ray source. Many typical high-power X-rays sources are minifocus, in the range of 1 mm across.This limits the resolution of images to that of the detector: a very fine detector is needed to get high resolution when no magnification is possible. Microfocus means the size of the X-ray source is only a few microns across (1 micron or 1µm is 1/1000th of one millimeter). With a microfocus source, geometric magnification is utilized with a standard detector can be used to gain an image with much higher spatial resolution.
More information is available by visiting http://www.nikonmetrology.com.