II.3. Radiography and Tomography for the Visualization of the Metal Wall and the Interior

A radiographic image is obtained by transmitting a beam of X-ray photons through an object. The transmitted X-rays are recorded by a detector placed at the rear of the object (fig. 382). Depending on the thickness of the object and the material from which it is made, the direct transmission of X-rays through the object is blocked, or attenuated, to a greater or lesser degree. Attenuation can be caused either by absorption or by scattering of the photons. Where X-rays pass through the object easily, the detector is strongly exposed, causing the image to appear dark. Where X-rays are more heavily attenuated, the image will appear light (only rarely is this convention inverted).

The actual attenuation of X-rays is dependent on two main parameters of the object being imaged. The type of atoms that compose the object (copper, lead, tin, iron, et cetera) greatly influences the attenuation. In general, the higher the average atomic number of the matter, the more attenuating it will be, so an object made of lead will be dramatically more attenuating that an object of the same dimensions made of aluminum.

The thickness of the object to be imaged also directly affects the attenuation of X-rays. Given two objects made of the same material, one twice as thick as the other, the thicker object will attenuate an X-ray beam twice as much as the thinner (fig. 383). Naturally, a porous object will attenuate X-rays less than a similar solid object, depending directly on the degree of porosity.

Open viewer Figure 382

Open viewer Figure 383

The attenuation of X-rays by an object is also dependent on the energy (or wavelength) of the X-rays used for imaging (fig. 384). Higher-energy X-rays are generally more penetrating than lower-energy X-rays. An X-ray tube may be set to produce X-rays at a range of energies. The tube is set to a certain energy level (expressed in kilovolts, or kV), which determines the maximum energy of X-rays produced (expressed in kiloelectron volts, or keV).

The nominal voltage setting may be somewhat misleading, since X-ray tubes produce a continuous spectrum of X-rays, with energies ranging from close to zero up to the nominal voltage setting. The dominant or average energy of the resulting X-ray beam will normally be between 25% and 50% of the nominal energy, with very little radiation actually being produced at the maximum level.

Open viewer Figure 384

Most radiography for cultural heritage is conducted with X-ray tubes. Common industrial X-ray tubes have maximum nominal voltages of 100–450 kV (rarely up to 750kV). Conventional industrial X-ray tubes suitable for studying bronze sculpture are heavy and require large auxiliary transformers as well as circulating liquid cooling systems, making them difficult or impossible to transport for use in situ. Increasingly, portable X-ray tubes that are air cooled and can be plugged into an ordinary electrical wall socket are becoming available.

Also becoming available are pulsed X-ray systems that are compact and battery powered, and generate multiple short pulses of X-radiation (measured in nanoseconds), with maximum voltages approaching 400 kV. The short pulse length significantly reduces the health and safety hazard associated with their use, and their small size makes them suitable for in situ imaging. Unfortunately, the low total photon output of these generators means that they can only be used effectively with digital radiography (DR) detectors, and not with film or computed radiography (CR) systems (see below).

In instances where conventional X-radiation is too limited in energy to penetrate high-density and/or thick artifacts (figs. 385, 386), gamma radiation may be used for imaging. Gamma rays are, like X-rays, photons of high-energy electromagnetic radiation, but are by definition produced by the decay of a radioactive source. For industrial gamma radiography, the two most common source materials are iridium-192 and cobalt-60. Radioactive sources have fixed, nearly monochromatic energy output. Iridium-192 has its primary gamma emission at around 380 keV, while cobalt-60 is at around 1250 keV. Iridium-192 allows bronze statues with thickness of up to 5 cm to be examined,³ while Cobalt-60 will reach 10 cm. Contrast is rarely as good as for conventional X-ray radiography (fig. 194). For very thick objects, an alternative to gamma radiography is to use high-energy X-rays (6000–30,000 keV) produced by linear accelerators or at larger particle accelerator facilities such as synchrotrons.

Open viewer Figure 385

Open viewer Figure 386

Open viewer Figure 194

When exposed to high-energy X-rays (300–400 keV), secondary electrons are emitted by some of the atoms present in the upper surface layers of an artifact. The heavier the atoms present (that is, the higher their atomic number), the more intense the electron emission. These electrons may be collected on a film held in close contact with the surface; this is called electron emission radiography (figs. 387, 388). X-ray film is much more sensitive to electron exposure than to high-energy X-rays, so the dominant image produced on the film will reflect the distribution of heavy elements on the surface.

As opposed to transmission X-radiography, only the upper 50–100 µm of surface will contribute to an electron emission radiograph, and areas of higher density will appear darker rather than lighter. Electron emission may be used to great effect in imaging inlays of different metals (particularly silver or gold) in flat or very low relief objects such as bronze plaques (fig. 389).

Intimate contact between the film and the surface is required in order to limit the absorption of the emitted electrons by air. Since the film is pressed onto the surface, it cannot be used on fragile surfaces. Unfortunately, the preferred (single-sided) industrial films are not produced anymore. CR plates are much less sensitive than film, and DR detectors are unsuitable for electron emission work.

Open viewer Figure 387

Open viewer Figure 388

Open viewer Figure 389

Currently, three main types of X-radiographic detectors are in common use: silver halide film, CR plates or sheets, and DR detectors.

Film offers a wide range of sensitivity and very high resolution, and it is very easy to cut and bend in order to adapt to the different shapes and sizes of the objects. Exposure times are typically on the order of several minutes, though exposures measured in hours are not unheard of for large sculptures. Processing of film radiographs is relatively time and labor intensive, taking about ten minutes to develop a single sheet. Silver halide film is becoming somewhat more difficult to obtain, though a wide range of silver halide films are still produced for industrial radiography. The chemicals used to process film are becoming increasingly subject to regulation as pollutants.

CR uses photostimulable storage phosphor plates to record images. These plates may be housed in rigid cassettes or used as loose, flexible sheets similar to traditional film. First the plates are exposed, and the phosphor layer captures the energy of the incident X-rays. The plate is then inserted into a scanner that scans a focused laser across the surface, stimulating the release of the stored energy as light. The intensity of the luminescence at each point on the plate is recorded by an optical sensor, which generates a digital image. Scanning can take between one and ten minutes, depending on the resolution settings and the size of the plate. Plates can be erased and reused hundreds of times if well cared for. The range of sensitivity of CR systems is at least as good as film, while the resolution is not quite as good, with pixel dimensions of around 25–50µm. The plates are available in many sizes, including long strips around 35 cm wide and 150 cm (or more) long. They can, if necessary, be cut and bent like film. Exposure times are typically a quarter or half as long as film. Modern CR systems usually incorporate digital filters for edge enhancement and other image optimization tools. These tools can dramatically improve the visualization of subtle features in radiographs.

DR systems use fixed arrays of X-ray-sensitive pixels that send image data directly to a computer through a wired or wireless interface, where it can be captured, viewed, and saved in real time. DR detectors have a wide range of sensitivity, but overall sensitivity is much higher than with film or CR, resulting in much shorter exposures, often only one or several seconds. Averaging of multiple exposures is typically recommended to reduce noise. The short time needed to generate radiographs makes it easier to produce many different views of an object and facilitates the acquisition of short movies or image sequences that aid in the interpretation of complex three-dimensional forms (figs. 121, 390, 498, 564). With sufficient rotational precision, image sequences suitable for full computed tomography (CT) can also be generated (see II.3§2 below).

DR systems yield images with lower resolution than film or CR, with pixel dimensions on the order of 100–200 µm. The detector plates are manufactured in fixed sizes (typically from 20 × 20 cm up to a maximum of about 40 × 40 cm) and are housed in rigid cassettes that cannot be cut or bent. Most modern DR systems also include the same kind of digital image optimization filters that are found in CR systems.

Some DR systems utilize line-scan sensors that require an x, y displacement system to allow the complete scanning of the object. These detectors are often used for X-ray tomography. Their main advantage is a much lower cost than array detectors. Some systems also enable the scanning of much larger surfaces (100 × 100 cm versus around 40 × 40 cm) (see table 15 Open viewer).

Open viewer Figure 121

Open viewer Figure 390

Open viewer Figure 498

Open viewer Figure 564

Correct image exposure is normally obtained by adjusting the exposure time (usually expressed in seconds or s) and the tube current (expressed in milliamps or mA).⁴ The current determines the rate of X-ray production by an X-ray tube, or its brightness. The product of the current and the exposure time (mAs) determines the total quantity of photons emitted and is thus commonly used as a designation of overall exposure. The distance between the X-ray source and the detector has a strong influence on the time and current necessary for a proper exposure. Total exposure is inversely proportional to the square of the distance, so doubling the distance between source and detector requires that the exposure time (or current) be increased fourfold.

Contrast in a radiographic image can be controlled by adjusting the voltage setting of the X-ray tube. The higher the voltage, the lower the image contrast will be. As a general rule of thumb, select a voltage that is just above the minimum energy required to effectively penetrate the most attenuating region of the object.

To guarantee maximum image sharpness, three main parameters may be adjusted: the distance from the detector to the object, the size of the X-ray tube’s focal spot (the area from which the X-rays are generated), and the distance from the focal spot to the object. The detector should be placed as close as possible to the area of interest of the examined object (fig. 391 3, 4), keeping the detector surface perpendicular to the beam (fig. 391 5, 6).

Very often the morphology of a three-dimensional subject does not allow close contact with all parts of it. Sharpness and depth of field can be increased by using a small focal spot. Depending on the model, the X-ray tubes may have different sizes of focal spots; the largest one allows the use of the entire power range of the tube, while with a smaller focal spot only the lower power range is accessible. The smaller the focal spot, the sharper the image (fig. 391 7, 8). Compensate for a consequent reduction in intensity by increasing exposure time.

To further increase sharpness, increase the distance between the X-ray source and the object (fig. 391 1, 2). This has the same effect on sharpness as shrinking the focal spot, and has the added effect of reducing geometric distortion. The effective exposure will be reduced according to the inverse square principle, so exposure time will increase accordingly.

Open viewer Figure 391

When making a radiograph, part of the X-ray beam is absorbed by the object and another part is transmitted to the detector to form the image. But at higher energies, a significant part of the beam is also scattered in all directions by the object itself and the room’s walls, floor, and ceiling. This generates fogging of the detector and a reduction in contrast (fig. 392). There are several strategies to reduce the effects of scatter. Lead shutters or a collimator may be placed in front of the X-ray source so that only the necessary area of the object is exposed by the beam, and X-rays that would fall outside the detector area are blocked. Lead sheeting several millimeters thick may also be placed immediately behind the detector and/or around the object to absorb the backscattered X-rays coming from the surroundings (walls, ceiling, floor).

Open viewer Figure 392

The primary X-ray beam may be filtered at the source to selectively attenuate lower-energy X-rays, and thus increase the average energy of the output spectrum. This is called beam hardening and is usually accomplished using aluminum or copper filters from one to several millimeters thick. Such filtration reduces overall X-ray output (requiring longer exposures) but can increase the effective penetration of the beam and also reduce scattering to a certain degree by attenuating non-penetrating X-rays that still contribute to scatter.

At high-energy exposures, silver halide film may benefit from being inserted directly between two thin lead screens (around 50 µm thick). High-energy X-rays cause the lead to emit electrons, which exposes the film very efficiently, thus reinforcing or intensifying the image. These screens are frequently used with film and CR detectors for bronze objects (figs. 37, 393), but are not useful for DR detectors.

Open viewer Figure 37

Open viewer Figure 393

Each sculpture is unique because of its shape, material, and facture. Thus, the geometric configuration of the X-ray source, the sculpture, and the detector must be optimized for each object, and unlike the case with medical or industrial applications, no preestablished protocol can or should be applied (compare for instance the operating conditions for three different bronzes as reported in figs. 32, 37, 394). The image obtained in an X-radiograph is the projection of a three-dimensional volume on a two-dimensional surface. The features present on many different planes are therefore merged, and determining the depth of any given feature can be very difficult.

To help locate features in three-dimensional space, it is therefore extremely helpful to make multiple images from different angles. A front view and a side view are the minimum for a radiographic study, while numerous views from different angles may be of great benefit. Complementary information will appear in the various images, leading to a clearer interpretation of the object. To aid in interpretation, it is very helpful to also have on hand high-resolution photographs taken from the same angle as the radiographs.

The speed of acquisition associated with newer DR systems allows the goal of multiple views overlaid with visible images to be more readily attainable. When combined with a digital turntable designed for product photography, it is even possible to produce radiograph sequences that capture a sculpture rotating through 360°. The angular interval between frames can be customized (5° is often satisfactory) so that an image set can be generated in around 20 minutes and can then be processed and viewed in a web browser. After radiography, the X-ray source may be removed and a digital camera may be placed in the same position allowing visible images and radiographs to be overlayed (fig. 498). If a simple easel is used (fig. 563), the detector can be repositioned, keeping the object and x-ray source fixed, so that multiple series of radiographs can be composited to yield image sets of objects much larger than the detector panel (fig. 564). This type of 360° X-ray image package provides substantially more information to the viewer than individual radiographs and yet is more accessible and simpler to produce than a full CT scan.

As a general rule, the larger the sculpture, the greater the optimal distance between the X-ray source and the object in order to minimize geometric distortion and enhance sharpness. For example, a distance of two meters is enough for a typical sculpture smaller than one meter. For a taller and broader sculpture, the X-ray tube might be positioned three meters from the object. If a complete radiograph of a medium-to-large sculpture is desired, limitations on detector size often make it necessary to make several successive exposures and stitch them together to obtain a composite radiograph. Depending on the file size of each radiograph, stitched images can generate files of several hundred megabytes, requiring computers with adequate processing power.

When images are going to be stitched, the X-ray source and sculpture should stay in the same relative position for all exposures. The film or detector should then be moved for each exposure in an overlapping grid within a single plane perpendicular to the X-ray beam, and as close to the sculpture as possible.

Open viewer Figure 32

Open viewer Figure 37

Open viewer Figure 394

Open viewer Figure 498

Open viewer Figure 563

Open viewer Figure 564

In theory, one should use the lowest voltage possible as long as the X-ray can penetrate the object being studied. If the material is thin or if the metal is corroded, 250 kV or below may be sufficient (fig. 394). In practice, for most bronze sculptures, the variation in thickness will be significant and the contrast of the resulting radiographs will tend to be very high. As a result, voltages exceeding 250 kV are frequently required, often with significant beam filtration in order to reduce contrast and yield proper exposure in all areas of the image (figs. 32, 395). Steps to limit scatter should also be taken.

A trade-off when using very high voltages with beam filtering is that local contrast within relatively homogenous regions of the radiograph will be very low, making subtle details difficult to resolve. Advanced image filters used in CR and DR can compensate for this difficulty to a considerable degree, but an alternative is to capture multiple images at lower voltage (higher contrast), exposing each to optimize exposure in different areas of the image. By using lower voltages on selected areas, small variations in thickness may be revealed. For objects with relatively thin walls, traces of hammering can be visualized by reducing the voltage (to less than 100 kV if possible), avoiding filtration, and working with long exposures and maximum tube current (fig. 396).

Open viewer Figure 394

Open viewer Figure 32

Open viewer Figure 395

Open viewer Figure 396

When making a tomographic reconstruction of a bronze sculpture, the subject should be positioned so as to present a minimum variation in thickness or density to the X-ray beam. It is often advantageous to angle the X-ray beam slightly upward or downward to avoid passing it directly through horizontal elements parallel to their long dimension. For example, outstretched arms or the tops of integrally cast bases can be very difficult to penetrate if the X-ray beam must pass through their length.

In tomography, the subject cannot come in contact with the sensor at any point in its rotation. This means that the bronze is often placed farther from the detector than in conventional radiography. Therefore, it is best to maintain a maximum distance from X-ray source to object as well as a small focal spot in order to maximize sharpness and minimize geometric distortions.

For conventional tomography, the bronze must appear entirely within the field of view of the detector on each projection to guarantee a satisfactory reconstruction. The size of the sensor (usually less than 50 × 50 cm for plane detectors) is thus a significant limiting factor for large works. It is possible to stitch together multiple radiographs into a series of larger projections, though this requires great precision and repeatability in the placement of the detector for each view.

In radiography, the parameters (voltage, intensity, exposure time) are adjusted for each shooting angle according to the thickness of the material to be X-rayed. In tomography, however, the shooting parameters cannot be changed during the scan. It is thus necessary to adjust the parameters to accommodate both the greatest and the least thickness encountered. To ensure penetration of thick parts and simultaneously avoid overexposure of thinner parts, very high voltage is often used and the beam is heavily filtered to further reduce contrast. Under these conditions, measures to reduce scatter are very important.