Dr. Fiona McNeill
The first X-ray image was taken by Professor Wilhelm Röntgen in 1895, a mere two weeks after he had first discovered the phenomenon he called “X-rays.” The remarkably obliging Frau Röntgen allowed her hand to be fixed between the X-ray source and a photographic plate: the image shows the clear pattern of the bones of her hand, the ghostly outline of the skin and muscles, and a ring that was studded with small jewels (Fig. 1). Röntgen recognized immediately the medical implications of his basic physics research, and the first medical use of X-ray imaging occurred three months later at Dartmouth College in Hanover, New Hampshire. Röntgen never applied for a patent, as he felt strongly that his discovery should be freely available. His generosity of spirit has resulted in millions of lives having been saved around the world. He could have been a millionaire, but instead died nearly bankrupt.
The “X-rays” that Röntgen discovered were, in fact, an electromagnetic wave. Radio waves, microwaves, light, and X-rays are all electromagnetic waves: X-rays are in the high-frequency (high-energy) part of the electromagnetic spectrum. X-rays are so useful to us for medical imaging because of the fortunate coincidence that our human bodies are made of distinct types of materials: bone and soft tissue. An X-ray image is possible because when X-rays pass through a material, they have a probability of interacting with the electrons in the atoms of the material. A denser material has more electrons per unit volume, so an X-ray is more likely to interact in dense material. In addition, elements with a higher atomic number have more electrons per atom, and so an interaction is more likely. In the case of humans, our soft tissue is made up of low atomic number elements such as hydrogen, carbon, oxygen, and nitrogen, and it has a density very similar to water. Bones are made from higher atomic number elements, such as calcium, and are denser. Higher density, higher atomic number bones therefore interact more strongly with X-rays than lower density low atomic number muscle and fat, so when X-rays pass through our bodies, more X-rays are absorbed or scattered in bone than in soft tissue (Fig. 2). This means that X-rays mostly pass through soft tissue, but are stopped by bone, so areas of soft tissue are bright on the X-ray image because the film is exposed. Areas of bone are dark on the image because fewer X-rays pass through and there is less exposure of the film. Broken bones, degraded bones from arthritis, and hairline fractures all show up readily in images because of the distinct contrast between bone and soft tissue.
The same principle, that different material types result in more and less X-ray interaction, can be used to study paintings. A common use of X-radiographs in art is to look for areas that have been painted over, or for canvases that have been reused. High atomic number materials attenuate X-rays strongly, that is, they allow few X-rays to pass through a material (Fig. 3). Commonly occurring high atomic number elements include metals such as lead and mercury. These are often found in the pigments used in oil paintings. If a single pigment was used in an even wash over the canvas, there would be no distinction on an X-ray image. All of the X-rays would be attenuated by the same amount, and the X-ray image would look flat and grey. However, paintings are often made with different pigments and are layered with different thicknesses of pigment. Sometimes, an artist will have painted on a canvas, and then decided either to reuse the entire canvas or change the image and paint over a section. This can sometimes (with luck!) mean that a pattern is observed on an X-ray image that is not seen by the naked eye. The X-ray shows the layers of pigment that are under the surface if (and only if) the combination of thickness, density, and atomic number of the pigment underneath the surface layer is different enough from other parts of the painting so that it attenuates the X-rays differently.
In The Unvarnished Truth: Exploring the Material History of Paintings, X-radiographs were used to investigate all of the paintings, and in the case of Van Gogh’s Untitled, Still Life: Ginger Pot and Onions, the X-radiographs were indeed different from the image observed with the naked eye. He had either reused the canvas, or repainted a section of the work. Fortunately for us, the pigment underneath was distinct enough in composition from the surface layer that it attenuated the X-rays, allowing the hidden image beneath to be detected. For further discussion on this, see the essay “The Analysis of Inorganic Pigments Using Spectrometric Techniques” in this volume.
In the case of Brouwer’s The Drinker / The Bitter Draught, materials behind the paint caused an interesting problem in the X-ray image. The Brouwer work is attached to a wooden cradle. When the X-ray was taken of the painting, the X-ray image (Fig. 4) was a really good picture of the cradle, rather than the painting itself. We were able to use an X-ray attenuation “trick” to reduce the impact of the cradle on the image by matching the attenuation of the cradle. Wood is a material that is mostly composed of the elements hydrogen, carbon, nitrogen, and oxygen. In X-ray terms, plastics can be considered to be very similar to wood because they are composed of hydrogen, carbon, and oxygen and have a very similar density. Elvacite is a product that consists of tiny (<1 mm) acrylic resin beads, and by using this it and pouring the beads into the back of the Brouwer painting, we could fill in the gaps in the cradle. The acrylic resin beads match the attenuation by the wooden cradle, and this means there is less contrast in the image between the cradle and the air (Fig. 5). We can therefore see the X-ray image (Fig. 6) produced from the layers of paint more clearly. The beads were poured back out after the X-ray, leaving the painting unharmed.
In modern medicine, most X-ray images are taken digitally, which means film is not used to record the X-rays so images can be taken and read immediately. There is no need to wait for film to be processed and images can be stored on a server for immediate access by the radiologist. However, for the studies of the paintings, film radiographs were used instead because the image resolution, which is how fine a spot or line size can be discriminated, was extremely important. Digital images are composed of a series of “dots” called pixels in a two-dimensional array. The resolution of the image depends on the total number of pixels in the image. A more blurred image is the consequence of fewer pixels in the array (Fig. 7). For our purposes, the digital X-ray scanners available to us did not have the high resolution we required. Their ability to detect fine detail, such as the lines in a canvas, was not good enough. However, films are still available that allow better resolution images to be taken than if using a camera. We ordered special high-resolution industrial X-ray film that was exposed by positioning the paintings face down on the film and taking the X-ray from above.
Using high-resolution film also allows us to view the threads on a canvas. The analysis of the pattern and density (per cm) of the weave of a canvas support of a painting, also referred to as thread counting, can provide important information regarding the piece. Research has shown that it is possible to compare the canvases of paintings attributed to the same artist and determine if they were cut from the same bolt.[i] The researcher will take a high-resolution X-ray film of a painting, scan it to a digital format, use a software algorithm to characterize the density and angles of the warp and weft pattern of the canvas, and then compare to a database of other paintings. The Thread Count Automation Project, co-directed by C. R. Johnson Jr. and D. H. Johnson, is the primary developer of this technique, which we applied to the Van Gogh painting. For a detailed discussion of this, see the entry on dendrochronology in this volume.
The principle of neutron radiography is very similar to X-radiography. A beam is passed through an object and a film (or digital camera) records the parts of the beam that are transmitted through the material. The McMaster Nuclear Reactor has a radiography beam port that is designed to allow imaging through objects using thermal neutrons, that is, low-energy (room temperature) neutrons. X-rays interact with the electrons around an atom, whereas neutrons interact with the nucleus of the atom itself. Because the interactions are based on different physical properties, X-rays and neutrons image different things. X-rays are excellent at imaging metal (because there are lots of densely packed electrons in metal), while neutrons are excellent at imaging materials containing hydrogen. Hydrogen has a high probably of interacting with thermal neutrons; in physics terms, we say hydrogen has a high thermal neutron capture cross section. There are other elements with high thermal neutron capture cross sections, two being cadmium and gadolinium. However, hydrogen is by far the most common element with a significant thermal neutron capture cross section, because it is an element in water, wax, and plastics, and therefore most radiographs are predominantly a result of interactions with hydrogen.
Combinations of X-radiographs and neutron radiographs can show complementary information. In the image of the lighter (Fig. 8), taken at the McMaster Nuclear Reactor by Nray Services Inc., the X-radiograph shows the metal striker and flint. The neutron radiograph shows the level of butane, which contains a lot of hydrogen, inside the lighter. In this project, Nray Services Inc. used neutron radiography to capture an image of Untitled by Rodchenko. We knew the image was painted on a spruce board. We used neutrons to image the painting for two reasons: we were interested in whether there was cadmium in the pigments; and because wood contains hydrogen, we hoped we could gain information about the wood. The resulting ghostly neutron image of this work showed the knots in the wood and a clear image of what looks like a panel board. It also showed an unfortunate crack in the centre of the painting (Fig. 9).
[i] A. G. Klein, D. H. Johnson, W. A. Sethares, H. Lee, C. R. Johnson Jr., and E. Hendriks, “Algorithms for Old Master Painting Canvas Thread Counting from X-rays,” Signals, Systems, and Computers: Proceedings of the 42nd Asilomar Conference (2008): 1229–33.