Paper published in Insight, July 2004
Keywords: Computed Radiography, Imaging Plates
Computed radiography is becoming more commonplace within the industrial NDT field. It offers several advantages such as defect recognition software, advanced analysis tools, faster exposure times, and lower energies - and yet many companies are still sceptical about its use. These companies are questioning whether they should wait for the new technology to improve before buying, or invest in the new technology now before being left behind. This paper attempts to answer some of the common objections raised against digital, and to compare the relative merits of analogue (film) radiography against the various types of digital. It also examines some of the new terminology surrounding computed radiography and highlights some of the differences in technical skills required from radiographers and analysts.
Much of the resistance to computed radiography is from radiographers themselves, who are used to using film and developers. Many are not comfortable with the new digital technologies, and are suspicious of the advantages offered by digital image processing. This paper attempts to dispel some of the myths regarding computed radiography, and takes an objective look at the various technologies on the market today.
TWI has investigated a range of materials of various thicknesses, using a variety of digital radiographic techniques. These techniques include Film scanning, Computed radiography using Imaging Plates (IPs), and Digital radiography using flat panel detectors. When the digital radiograph is created using an imaging plate, the process is known as Computed Radiography (CR). The resultant image is still a digital radiograph.
The materials investigated by TWI include:
- 1-60mm Magnesium castings in the Magcast project
- 75mm steel in the Rail Inspect project
- Aerospace composite samples in the Nanoscan project.
All of these projects are €2M European funded CRAFT projects. TWI is also engaged in several joint industry projects which compare digital to film radiography. This paper draws on the collective experience of these projects, and tries to answer the following questions:
- What are the different options?
- What are the advantages and disadvantages of the different technologies?
- How do they compare to film?
- What objections are typically raised by radiographers?
- What new skills will they have to learn?
1.1 Potential benefits
The potential benefits of digital radiography include  :
- Sharing information
- Digital enhancement
- Automated defect recognition
- Ease of automation
- No consumables, and no chemicals
- Faster exposure times
- Greater linearity and range.
1.2 The digital format
The digital image is constructed by a matrix of square pixels; a typical digital radiograph may be comprised of many tens of Mega-pixels. This is also typical of digital photography, and indeed, digital radiography shares many similarities with digital photography. Both have a limited number of pixels, which are typically much larger than the grains in film - this can often lead to problems related to resolution and image sharpness. This problem can be overcome using projection magnification with a mini or micro-focus set. Digital radiography is also generally faster than radiographic film, just as digital camera chips are faster than photographic film. This has obvious health benefits for radiographers using the digital format.
Black and white digital photographs are normally constructed of 256 shades of grey alone. Eight bits are typically used, as the human eye cannot distinguish any smaller difference in shade. Digital radiography is normally constructed of 12 or 16 bit images that mean that, in theory, they contain more information than the human eye is capable of seeing. This extra information can be revealed by simple digital image processing, typically by adjusting the brightness and contrast.
Compression is usually used in digital photography to reduce the size of the files, but can lead to the generation of artificial artefacts that could be mistaken for flaws. Figure 1 shows two digital images. The first image has been subjected to a typical compression value, while the second has been overly compressed. A closer examination of the first image reveals errant pixels, particularly around the tip of the beak. Digital radiographs are typically left uncompressed for this very reason. The combination of large pixel numbers, high bit rates and zero compression, leads to extremely large file sizes when dealing with digital radiographs.
Fig. 1. Compression of digital images
For reporting purposes digital radiographs are often adjusted for brightness and contrast then exported as jpegs. These jpegs can then be cropped and resample before being pasted in to a report or paper. This makes communication of radiographic data much easier.
1.3 Creating a digital radiograph
There are three main technologies used to create a digital radiograph. These are:
- High resolution scanning of film
- Phosphor Imaging Plates (IPs)
- Flat panel detectors.
Each has their advantages and disadvantages, which will be discussed in turn. It should be noted that the digital radiographs created using the phosphor imaging plates, described in the next section, are typically referred to as 'Computed Radiography', while the flat panel detectors are described as 'Digital Radiography'. The only difference between the two is the method of creation. Once the image is displayed on screen there is absolutely no difference between the two.
2. High resolution scanning of film
This method is used when a digital image is required, but a traditional film has been used to capture the radiographic image. There are several reasons why this may be desirable. The most obvious reasons include archiving and communication, but there are more technical reasons why this may be needed. A digital radiograph of a low-density component may contain much more information than the human eye is capable of seeing. This is also true with a film radiograph, although the effect will not be linear. This information may still be valuable and can be extracted by digitising the film much in the same manner as a paper document can be 'scanned' in to a PC using a flat bed scanner. There are several service providers in the UK who can digitise films. The results are typically burnt on to a CD and posted back to the client. The pixel-pitch (see section 5.1) of such scans may vary, but is typically around 25-50µm.
3. Imaging plates
The use of phosphor Imaging Plates (IP) to capture a digital radiograph is normally referred to as 'Computed Radiography'. The imaging plate itself is a flexible white plastic sheet that is used in much the same way as a film. It is placed in a light proof cassette with lead screens as normal, then placed in exactly the same position as you would film - it is even possible to bend the flexible plate around a pipe. The exposure time is then calculated. The typical exposure dose is approximately 70% of the KVs used for an Agfa D7, and 10% of the exposure, for lower KVs. For higher KVs it may be closer to 50%. The current plates on the market will give a similar radiographic quality of a D7, but newer plates on the market have the potential of delivering quality levels of a D4/D5. 
Once the plate has been exposed it is removed from its light proof cassette in a darkened room. There is no need for a 'dark room' as such. The plate is wrapped around a drum, which is then placed in the scanner ready for digitisation, as shown in Figure 2.
Fig. 2. The Imaging Plate Scanner & Imaging Plate (IP)
The scanning procedure can take between three and twelve minutes depending on the size of the plate and the required resolution.
Very briefly, the latent image is created when X-ray quanta hit the grain structure of the plate. The energy is absorbed by the atom, without creating a visible image. This energy is then released in the scanner by a He-Ne laser.  This laser releases a quantity of blue light, which is detected by a Photo Multiplier Tube (PMT).  The amount of blue light is a linear measure of radiographic 'density' at this point. The typical pixel-pitch of such scanners is 50-150µm. The voltage across the PMT can be varied as 'gain'. A high value of gain during scanning will lead to a darker radiograph, which can be used to reduce exposure times. This will of course lead to lower sensitivity. A lower value of gain will result in a lighter image. In order to achieve the required greyscale value it would be necessary to expose the plate for a longer period, which would increase the sensitivity.
The latent image can be erased by placing the imaging plate on a normal light box for five minutes. Once the image has been fully erased the plate can be reused. In general the plates are very robust and are ideal for industrial purposes where a more delicate sensor may be damaged or adversely affected by dirt. The plates can be cleaned with Isoproponal cleaning fluid to remove dirt and grease from handling; grease from hands can severely affect the final image of the radiograph. The manufacturers claim that the plates have a theoretical working life of many tens of thousands of exposures, but industrial users have claimed that the practical working life is more likely to be many hundreds of exposures, as the plates can become scratched.
The plates are particularly sensitive to lower energies, as they were originally designed for the dental and medical field, and have been adapted for industrial NDT. This sensitivity to low energies is advantageous, but there can be problems with scatter in thicker walled objects. The effective use of filters and screens is essential.
4. Flat panel detectors
There are two forms of flat panel detectors - Amorphous Silicon, and the newer more expensive Amorphous Selenium. Both offer lower exposure times.  Figure 3 is a cut away view of Agfa's Amorphous Selenium flat panel detector. These panels are capable of achieving film like quality, but with a typical pixel-pitch of 100µm. For a direct conversion flat panel a photoconductor is used instead of a scintillator X-rays are converted directly into charge carriers. 
Fig. 3. An Amorphous Selenium flat panel detector Courtesy of Agfa
The panel is connected directly to a PC for power and control. This enables the system to be used in real-time, but can make site work difficult, as the panels are often delicate. There are flat panel detectors on the market that are protected in an industrial enclosure with their own PC card and battery power supply for site work, but panels are more normally used in automated production lines.
The average lifetime of these panels should be taken into account when considering their use. Overexposure, too higher energies, and failure to allow the panels to 'rest' between exposures can severally decrease their effective life span. The price and specification should also be taken into consideration when considering the use of these panels. In particular, the pixel-pitch of 100-150m can be a problem.  This may be overcome by projection magnification, but this will be at the loss of imaging area. However, the price and specification of these panels are improving rapidly.  When considering purchasing such a panel it is important to note that the cost and quality of these panels varies tremendously.
5. Pixel-pitch and density
Pixel-pitch is the distance between pixels, and should not be confused with resolution. In Figure 4 the eye of the parrot on the left hand side has been magnified. It takes several pixels to image an artefact correctly. When sampling a waveform the Nyquist frequency of two times the maximum frequency contained within the waveform is used to derive the minimum acceptable sampling frequency. However it is generally agreed that a sampling frequency of two to four times the highest frequency is more appropriate. The same is true when sampling digital images. Digital radiography, with its inherently low resolution, may be ideal for defects such as porosity, but may fail to correctly image planer faults such as cracks, without such a technique as projection magnification. When using digital it's beneficial to check the theoretical unsharpness. If the unsharpness is greater than the pixel-pitch then the unsharpness will be the limiting factor in resolution and visa-versa.
Fig. 4. Demonstration of pixel-pitch
The density of a film radiograph is based on the logarithmic ratio of the incident light from the viewer and the transmitted light. That is to say a radiograph with a density of 4 reduces the light intensity by a factor of 10,000. Obviously a digital radiograph is not placed on a viewer, so the same evaluation of radiographic lightness and darkness cannot be used. Digital radiographs typically use greyscale values instead. A 12 bit image will have a range of greyscale values from 0 to 4095. Zero being black and 4095 being white. These numbers are hardly convenient, and the situation becomes even more unwieldy when using 16 bit. The author suggests that a simple percentage scale be employed - 0% representing a completely white pixel, and 100% representing a black pixel. In this manner the higher values of percentiles represents a darker image, as is consistent with radiographic density. The values would not be affected by the number of bits.
6. Comparison with film
Table 1 compares film, imaging plates and flat panels with such properties as IQI sensitivity and pixel-pitch. Film and Amorphous Selenium are capable of achieving high sensitivity over a range of energies. Film has a far greater resolution than flat panels, while the flat panels require a much lower exposure. This is also seen in digital photography, where a traditional film camera is far more able to capture sharper images, but requires far more light to be able to do this. Imaging plates have the least sensitivity, but have the best resolution of the various digital techniques.
Table 1 Comparison with film - 1
|Pixel pitch (µm)
Table 2 details some of the other advantages and disadvantages of film, imaging plates and flat panel detectors.
Table 2 Comparison with film - 2
|Bend around pipe
|Ease of automation
Figure 5 shows the equivalent Penetrameter sensitivity for flat panels (digital radiography - DR), imaging plates (CR phosphor) and film.  Both imaging plates and flat panels achieve the required 2% IQI sensitivity in a relatively short time in comparison to film, at low KVs. Flat panels are able to achieve film like quality in relatively little time, whereas the sensitivity of imaging plates is limited. However Agfa are in the process of producing more sensitive plates.
Fig. 5. Equivalent Penetrameter sensitivity Courtesy of Agfa
Table 3 details the current standards open for public comment at the time of writing. It is believed that the ASTM standards are very similar.
Table 3 Current standards
|CEN/TC 138 N 540
||Non-destructive testing - Industrial computed radiography with phosphor imaging plates - Part 1: Classification of systems
|CEN/TC 138 N 541
||Non-destructive testing - Industrial computed radiography with phosphor imaging plates - Part 2: General principles for examination of metallic materials using X-rays and gamma rays
8. Analysis of the digital image
When viewing a radiograph on screen it is important to note that there may be far more information within the image than the human eye is capable of seeing, and that the monitor screen is capable of displaying. A typical monitor screen is made up of 1024 by 768 pixels, as shown in section 1.2.1, but the radiograph is likely to contain even more pixels than this, making it impossible to display all the image's pixels on the screen at one time without the use of an extremely high resolution monitor. Such monitors do exist, but again the human eye is also limited in resolution. When analysing a digital radiograph the software's 'zoom' function should be set to a 1:1 ratio to ensure that one pixel on the monitor is displaying one pixel of the image. Magnifying the image to a larger size than this would be acceptable, but shrinking the image to less than a 1:1 size, would be insufficient, as not all the image's pixels would be on display.
Most monitors are only capable of displaying 8 bits of black and white 'greyscale'. This is because the human eye is generally not capable of distinguishing any smaller difference. There are more expensive monitors on the market that are capable of displaying more than 8 bits, but their use is not essential, as the human eye may not be able to appreciate the difference. A graphics card capable of dealing with 12 to 16 bit greyscale images is essential however.
The full range of shades can be seen by adjusting the brightness and contrast, even if a lower quality monitor is used. Figure 6 is a radiograph of a 1-10mm magnesium stepwedge taken on an Imaging Plate. The single 16 bit image has been separated into three 8 bit images by adjusting the brightness and contrast of the original image, then exporting as a jpeg. Many radiographers and technicians are suspicious of changes made when magnifying the image or by simply adjusting the brightness and contrast, as they believe that this is 'manipulating' the image, which could lead to 'false' information being displayed on screen. This is not the case when using such simple functions as zoom, brightness and contrast and sharpening - in this case you are actually removing information to allow the remaining information to be seen clearly. However, some processes do genuinely distort the image - these functions include embossing and some edge enhancements. It is important to understand which functions in your software change the image, and which merely adjust the image to make viewing easier.
Fig. 6. A magnesium stepwedge at three levels of brightness
When adjusting the size, brightness and contrast for optimal viewing it may be beneficial to use your IQI wires or Penetrameter. Common sense would dictate that any adjustment that reveals more wires or holes would be better than one that doesn't. Several critics of digital radiography have suggested that both wire/Penetrameter IQIs and duplex wires be used simultaneously. One for sensitivity and one for resolution, as the resolution of digital images can be suspect due to the limited pixel-pitch.
8.4 Quantifiable measurements
One of the advantages of digital radiographs is that exact quantifiable measurements can be taken. In Figure 7 a simple line measurement has been taken. The operator has drawn a line across a stepwedge, and the software has measured the greyscale value of every single pixel along the line. This is an extremely simple example of what measurements can be obtained.
Fig. 7. Example of a line measurement Courtesy of CIT
8.5 Other functions
There are various other functions and filters that can be applied to digital radiographs. These tools can be used to aid defect detection or to reveal hidden detail within the image. Figure 8 shows three images of a valve. The first is the original unfiltered image. The second is the same image after sharpening, while the third has been embossed in the Region Of Interest (ROI).
Fig. 8. Original, sharpened and embossed image Courtesy of Agfa
8.6 Automatic defect recognition
Automatic defect recognition is used to analyse images for defects without the intervention of a human operator. A high degree of repeatability is required for automatic defect recognition, which means that it is usually employed on production lines using flat panel detectors.  The source is typically stabilised in some manner to provide consistent doses of radiation for each component examined.
Each image is subtracted from a 'golden image' of a perfect component. The algorithm then examines pre marked 'danger areas' where faults are suspected. Small errors in aligning the two images can lead to errors and false positives. The two images are first 'registered' to ensure that they are correctly aligned pixel to pixel before subtraction. This technique is often used in castings such as automotive aluminium wheels.
9. Computed tomography
Computed tomography is used to recreate three dimensional views or 'slices' through an object. Figure 9 shows a photograph of a radiographic system used to create tomographs. Essentially an object is placed on a turntable, which is rotated in a series of incremental steps, taking a digital radiograph, with a flat panel, at each step. These images can then be combined either as simple 'frames' to create a movie of the rotating component, or can be processed by a more complex algorithm to produce a three dimensional virtual object as shown in Figure 9. This image can be rotated to view from any angle, or can be 'sliced' to reveal new details within the object.
Fig. 9. Computed Tomography Courtesy of Agfa
10. Radiographers and Technicians
Technicians involved in traditional film radiography have not welcomed the advances made in digital radiography, and have often shown a deep mistrust of the new technology. This mistrust and scepticism was also apparent when photographic digital cameras were first introduced on to the market, and yet today there are very few photographers who do not own one. Many radiographers do not trust the new technology, and are particularly sceptical about the benefits offered, such as greater linearity, and even some of the simplest tools such as brightness and contrast.
The skills required to create digital and film radiographs are very different, as digital and computed radiography depends upon a high degree of computer literacy. The new skills required from radiographers include image processing, multi-media, networking and filing/archiving. There is a very real risk of losing experienced personnel if radiographers fail to adapt to the new technology.
There are currently three methods in which a digital radiograph can be created. The first is to take a traditional film radiograph and to scan it into a PC in much the same way as a flatbed scanner can be used to scan documents and photographs. The second technique uses phosphor coated Imaging Plates (IPs) to create the digital radiograph. This process is known as Computed Radiography (CR). The final technique employs flat panel detectors, which capture the image directly.
- If your image requires a fine-grain/high resolution film, the present digital sensors may lack the required resolution, for example with fine cracking; but keep an eye on the market as they are improving daily.
- If your application requires a medium-grain film and you want to lower exposure doses, you require a robust system and/or to be able to bend around pipes - consider Imaging Plates.
- Finally if your application is for flat items or castings, and you require a high throughput, full automation or real-time, consider flat panel.
In general digital radiography appears to be lagging behind digital photography by five to ten years. Five to ten years ago people asked whether digital cameras would ever take off...
The author would like to thank Eric Deprins of Agfa for his assistance and technical expertise.
- SK Bansal et al, 'Digital radiography vs conventional radiography - a comparison along with its image quality and benefits', Journal of medical physics 26 3 (2001)
- E Deprins, Agfa Antwerp, 'By personal communication' 2003
- Fuji 'Computed radiography technical review No14 Imaging plate', www.fujimed.com/sub/tech_review_14.pdf.
- E Samei, 'An experimental comparison of detector performance for computed radiographic systems', Med. Phys. 29 (4) April 2002
- GA Mohr and C Bueno, 'GE A-Si flat panel detector performance in industrial digital radiography', BINDT Insight Vol 44 No 10 October 2002
- P Willems, 'Image quality comparison of digital radiographic system for NDT', www.ndt.net
- M Purschke, 'IQI sensitivity and applications of flat-panel detectors and X-ray image intensifiers - a comparison', Insight Vol 44 No 10 October 2002
- SR Amendolia et al, 'Comparison of imaging properties of several digital radiographic systems', Nuclear instruments and methods in physics research A466 (2001) 95-98
- Agfa official web site ndt.agfa.com/BU/NDT/index.nsf/EN/radviewdigitalsystems.htm
- AG Vincent et al, 'Defect detection in industrial casting components using digital X-ray radiography', BINDT Insight Vol 44 No 10 October 2002.