COMPUTED RADIOGRAPHY THE STATE OF THE ART

2007 International Nuclear Atlantic Conference - INAC 2007 Santos, SP, Brazil, September 30 to October 5, 2007 ASSOCIAÇÃO BRASILEIRA DE ENERGIA NUCLEAR - ABEN ISBN: 978-85-99141-02-1 COMPUTED RADIOGRAPHY THE STATE OF THE ART Carla Alves Marinho 1, José Maurício Rabello 2, Marcos Aiub 3, Eduardo Iguchi 4, Ricardo Lopes 5 and Davi de Oliveira 6 1 Centro de Pesquisas e Desenvolvimento Leopoldo A. Miguez de Melo CENPES - PETROBRAS Av Horácio Macedo, 950, Cidade Universitária 21941-915 Rio de Janeiro, RJ carlamarinho@petrobras.com.br 2,3,4 Serviço de Qualificação Industrial - SEQUI -Petrobras Rodovia Presidente Dutra, km143 Bairro Vista Verde São José dos Campos, SP mauriciorabello@petrobras.com.br aiub@petrobras.com.br iguchi.fbts@petrobras.com.br 5,6 Laboratório de Instrumentação Nuclear (LIN / COPPE) Bloco I sala I-133 Universidade Federal do Rio de Janeiro 21941-914 Rio de Janeiro, RJ ricardo@lin.ufrj.br davi@lin.ufrj.br ABSTRACT Despite the advantages obtained with the radiographic inspection, the risks inherent to the exposure of ionizing radiations not always represented the major limiters regarding the employment of the technique. However, with the arrival of the computed radiography, the industrial radiology gets space again due to the benefits provided by the new technology, among which the following are outstanding: possibility of reduction of the time of radiography exposure and, consequently, of the marking area, direct acquisition of images, reduction of consumables and chemical waste and possibility of using again the detectors (imaging plates). Before such a favorable scenario, the computed radiography is being highlighted as the solution for almost all the problems, even without the full knowledge of its advantages and restrictions. This work briefly presents the state of the art related with the industrial applications (in-service inspection of wall loss and weld inspection) and the limitations noticed up to the moment are discussed. 1. INTRODUCTION This work aims at presenting, in the most accessible form possible, a summary about the computed radiography technique and experiences acquired with its use to inspect welds and monitor corrosion. Petrobras, represented by the Research Center CENPES and by the Engineering Department, has been developing an intense work in the study and technical evaluation, in association with companies, universities and national and foreign institutions. 1

2.1. The Technique General View 2. BIBLIOGRAPHIC REVIEW The computed radiography technique (CR), despite keeping the same physical principles of the conventional technique, presents peculiar characteristics that must be considered upon its application. Leaving the use of X-ray films aside, CR uses photostimulable X-ray storage phosphors plates, or imaging plates, which, by absorption of ionizing radiation, form a dose proportional latent image. During the radiation incidence, electrons and holes are generated and captured locally in storage centers: F-centers or traps. The stored information is read out by scanning with a focussed red laser. This leads to a local excitation of the trapped electrons which subsequently recombine with the trapped holes causing the emission of light. This emission is known as photostimulated luminescence (PSL). The information is then recorded by means of a photomultiplier, digitized by an A/D converter and displayed with the aid of a computer [1, 2]. Figure 1 shows the process. (a) (b) (c) Figure 1. Computed radiography: (a) Radiation incidence: electrons and holes are generated and captured in F-Centers or traps ; (b) Read out by scanning with a focussed red laser; (c)the whole process of image generation. Other models were proposed to describe the phenomenon, but the discussion about it is out of the scope of this work [3,4]. The remaining information in the imaging plate will then be erased by a strong white light, so the plate can be reused. INAC 2007, Santos, SP, Brazil. 2

2.2. Computed Radiography System A Computed Radiography System (CR System) is a complete system of a storage phosphor imaging plate (IP), a corresponding read-out unit (scanner) and a system software, which converts the information of the IP into a digital image [5]. The IP and scanner type, together with the exposure conditions and the scanning parameters, results in a CR system class. Any change in one of these two components (another IP or scanner type) creates a new system. -Imaging Plates (IP) Imaging plates, like conventional films, show differences in speed at defined radiation quality and resolution. An IP consists of a flexible polymer support which is coated with the sensitive layer. On top it is covered with a thin transparent protective layer. The sensitive layer of the most common systems consists of a mixture of BaFBr doped with Eu 2+ [2,7]. Resolution and image noise or homogeneity are determined by the structure and thickness of the phosphor layer; phosphor grains size is also an important factor [1]. Standard IP Types (ST) are thick, requiring less exposure time and generating images of lower quality due to the photo stimulated luminescence scattering along the thickness plate. The fine plates have a High Resolution (HR) due to the reduced dispersion, by they require an increase of the exposure time due to the lower efficiency to absorb the photons of the reflected radiation; consequently, the resolution in contrast is poor. The blue plates are outstanding as high quality plates. Figure 2 shows the results obtained by Ewert et al [7] using the same scanner model and radiographic parameters (current, distance between detecting source and current in filament), but upon varying the quality of the plates, with three different types. The images refer to a duplex wire IQI, that was better solved by the blue type, for high resolution applications. ST 1 Seifert Isovolt 320, 100kV, 1mA, small focus, SDD=1m ST 2 HR blue Scanner model X Pixel size 28,2µm (900dpi) Figure 2. Results obtained for different plates using the same scanner model and radiographic parameters. INAC 2007, Santos, SP, Brazil. 3

It is important to emphasize that the blue HR plate was 10 to 100 times slower than the ST1 plate, in accordance with the comparison bases between the signal-to-noise ratio of the images [7]. Another point that must be considered refers to the characteristics of the radiographic scanner, as the quality of the image strongly depends on the way it was acquired first. The quality of the phosphorous plates is also a variable extremely important as already mentioned before. With regard to the quality of the scanner, three parameters are relevant: laser spot size, pixel size and dynamic range. The equipment that presents a laser spot size with small dimensions shall provide a better spatial resolution than another that presents a larger reader, under the same conditions. The pixel size has an effect analogous to the laser spot size: the smaller the dimension, the better shall be the spacial resolution of the image. However, this statement is valid just up to the point in which the pixel size is equal or superior to the laser spot size; from the moment it becomes inferior, there is no gain. Figure 3 provides the scheme of these two parameters and the relationship between them. It is interesting to emphasize that the reduction of the pixel size increases the scanning time, that is, the time the scanner takes to read a plate, what affects the total time of the radiographic inspection. b.1 b.2 P P b.3 P (a) (b) P Tamanho Tamanho do do pixel leitor ótico Figure 3. Scanner parameters: (a) Different laser spot sizes; smaller dimensions, better resolution; (b)relationship between laser spot size and pixel size. b.2 will show better resolution than b.1. It can be seen that b.3 shows read-out overlaping. The dynamic range of the equipment must also be noticed. It is related to the quantization of the digital image, expressed in number of bits for the resolution in grey levels. This property limits the smallest detail resolved in the direction of the reflection of the radiation bundle [8]. The available systems of phosphor imaging plates and corresponding laser scanners cover radiation dose differences up to 10³-10 4 in a linear scale. This feature reduces the number of exposures for objects with high wall thickness difference. The system provides more grey levels than the human eye can distinguish, what, however, shows itself as a benefit during image processing [7]. The image processing system is the program used as the communication interface between the scanner and the user during the plate reading stage, as well as during the analysis/ treatment of the images stored in the computer. The resources required for the adjustment of visualization of images on the screen, dimensional measurements and profiles traces in gray levels, for example, are made available by these programs, which are proprietary and exclusive software for each manufacturer of plates and radiographic scanners. There are several other programs able to process digital images, besides the proprietary ones. One example of these programs is the MatLab, which allows several image processing operations, allowing, inclusively, the creation of proper processing programs [9]. INAC 2007, Santos, SP, Brazil. 4

However, it is important that the user aims at working with filing formats, by the time of the exportation of images that do not cause loss of data, so that the analysis is not compromised. Another important variable for a computed radiography system is the monitor used to view and report of the radiographic images. In Petrobras internal standards the monitor is considered as an integral part of the radiography system and they provide its minimum requirements: minimum luminance of 100 cd/m 2, resolution equal or better than 1280 x 1024 pixels, maximum pixel pitch of 300 µm and luminance ratio (maximum luminance/ minimum luminance) equal or better than 100:1. 2.3. Evaluation of the Quality of the Image The evaluation of the quality of the image obtained with the conventional radiography is based on the control of the optical density and radiographic sensitivity. The project code of the equipment defines the optical density range and the minimum radiographic sensitivity, which are checked through the use of densitometers and through the use of IQIs, respectively. In the case of the computed radiography, there still remains the need to use IQIs, of contrast and unsharpness; however, there are quality parameters that can not be neglected: basic spatial resolution and signalto-noise ratio (SNR) [5,7]. They are new concepts that shall make part of the culture of those persons that want to work with the technique and get the best results from it. Basic Spatial Resolution (SR b ) The basic spatial resolution (in µm) is the effective pixel size of a CR system, being the minor detail resolved perpendicularly to the direction of the reflection of the radiation bundle, and can be measured in several manners. In normalization committees it is recommended the use of a duplex wire IQI method due to its simplicity. (EN- 462-5, E 2002) [5,7,10]. These standards provide the total unsharpness value (µt), in µm, which is related with the spatial resolution. SR b = ut 2 (1) The SR b is related to the size and intensity of the stimulating laser and imaging plate. The quite narrow laser beam stimulates a small area in IP, producing, thus, a small and well defined pixel (see Figure 3) [9]. However, the resolution does not depend exclusively on the dimension of the stimulating bundle. It also depends on the imaging plate. Figure 2 showed how the spatial SR b of a system can be improved through the use of better quality IPs. The STI plate reached 130µm versus the inferior value at 50µm obtained by the blue plate. It is important that the spatial resolution is defined in both directions of the IP X and Y, so that the worst value, i.e., the highest one, represents the system [5]. During the reading of an imaging plate, the scanning is performed line by line, and at the end of each line, the laser bundle returns to the initial position. As the imaging plate is moved simultaneously, this displacement is adjusted so as the next line to be scanned by the laser begins with a spacing that assures equal sampling in the X and Y directions of the plate. The scanning parallel to the displacement of the bundle is called fast scan, while in the perpendicular direction (or direction of plate s displacement) it is called slow scan. Both usually INAC 2007, Santos, SP, Brazil. 5

represent different values for SR b, and the best results, that is, the lowest values, generally occur to the slow scan direction. Signal-to-Noise Ratio (SNR) The SNR can be described as the quotient of mean value of the linearised signal intensity and standard deviation of the noise at this signal intensity. It depends on the radiation dose and the CR system properties [5]. The signal/noise ratios of industrial X-ray film systems are indirectly given in EN 584-1, E 1815, K 7627, ISO 11699-1 (table 1)[7]. Table 1. X-ray film systems according to different standards and their corresponding SNR and G2/σD values. Film systems are characterised by the gradient G D and the granularity or noise, σ D, both of them at D = 2. The most important parameter for the perception of fine flaws is the gradient over granularity ratio G 2 /σ D, which can be used to calculate the corresponding SNR [7]. The conversion of G 2 /σ D into SNR values is based on the assumption that both systems, NDT film system and digital detector systems, provide signals (optical Densit or photo stimulated luminescence), which are approximately proportional to the exposure dose. The SNR can be calculated for linear systems by [7]: SNR = (G 2 /σ D ) / ln(10) (2) It is noticed that the values presented in table 1 for SNR were based on [5] and [10] in the definition of the CR system classes. The European standard [5] promoted the division into six classes (Ip1 through IP6), exactly in accordance with the lines of the last column of table 1, and class IP1 corresponds to the highest value of SNR (130) and the IP6 to the smallest value (43). Alternatively, the US standard (3) created four classes: special, I, II and III, adopting as basis values equal to 130, 65, 52 and 43, respectively. The SNR values of film are measured with a circular diaphragm of 100 µm diameter after exposure to a diffuse optical density of 2. The diaphragm area has to be converted into a square shaped area for comparison of film to digital images or detectors. The equivalent square of a picture element (pixel) amounts to 88.6 x 88.6 µm [7]. Therefore, the standards (CEN, ASTM and Petrobras) for CR require minimum normalised SNR N limits. The measured SNR meas has to be corrected by: INAC 2007, Santos, SP, Brazil. 6

SNR N = SNR meas.88,6/sr b (3) This means to say that for the characterization of systems and analysis of the quality of images of the values obtained for SNRn, there shall be compared the SNR values of table 1. A last interesting observation is that with the use of high quality plates (HR), the SNR values obtained decrease, being off set just with the increase of the dose, or exposure. Of course this effect continues up to a certain point, and there is a moment from which the increase of dose would not be good. Anyway, this means increase of the exposure time upon using high performance systems. Figure 4 illustrates this discussion. SNR must also be defined in both directions of the plate, as SRb. Most proprietary software still does not provide this value immediately to the user, from a certain measurement window opened on the screen, for example, making available just the values found for the linearised signal intensity (I), point to point, expressed as grey levels. But, through the knowledge of the intensity values of the pixels of a specific region of the image, the SNR can de determined (figure 5) [5, 9, 10]. ST 1 ST 2 HR blue Figure 4. SNR values for different IP types under the same exposure conditions. The SNR is obtained as follows Where: Figure 5. Definition of SNR based on point to point intensity values, where i and j are respectively the number of lines and columns of the area used to define the SNR and Xi,j are the intensity values of each pixel. INAC 2007, Santos, SP, Brazil. 7

This calculation method is time consuming and hard. Due to this fact, in order to help user, the European standard [5] lists the intensity values (I) to the corresponding normalized SNR (SNR N ). The detailing of the procedures to get the SNR or I values does not make part of this scope. The German Federal Institute for Material Research and Tests BAM (Bundesanstalt für Materialforschung und Prüfung) makes the executable program ISee! available for free (version 07.01 demo), which has a tool quite useful, among others, through which the SNR N of the radiographic image can be easily obtained. The user informs the basic spatial resolution and the program calculates the SNR N. In short, the CR systems shall be characterized by the highest SR b value and by the smallest SNR N value [5,10]. Now, these are the parameters of interest that must be evaluated to analyze the quality of a digital radiography, inclusively to compare with the X-Ray NDT films. 3. FILM REPLACEMENT BY IMAGING PLATES NORMALIZATION The digital detectors technology were basically developed for medical applications, allowing a quick acquisition of radiographic images, with dynamic range larger than the industrial films, and presenting, however, and in the majority of the cases, low spatial resolution in comparison with the films applied in NDT [7, 11]. It is required to notice that, mainly when it is aims at using the technique for inspecting welds. Unfortunately, geometrical unsharpness values are not harmonized world wide. European standards define the geometric unsharpness as a function of the wall thickness for two classes (standard A and enhanced B). ASTM and ASME standards require quite moderate unsharpness values, especially in the lower wall thickness range. The original idea was to require about the same unsharpness as the perceptible wall thickness contrast. These values are usually between 1% and 2 % of the maximum material thickness. The typical NDT testing sensitivity in USA requires the 2-2T penetrameter perceptibility, which even allows an unsharpness of 4% of the material thickness [7]. Besides the facts exposed above, the standards for the computed radiography have been published as from 2005, and there are already some controversial points between them. The mandatory appendix of ASME V, which deals with the computed technique, just make requirements about the control of sensitivity with the use of IQIs, what showed itself insufficient to get radiographies of quality. The European standards EN-14784, parts 1 and 2, as well as the US standard ASTM-2446, make requirements as to the values of signal-to-noise ratio (SNR) and spatial basic resolution, besides the use of the IQIs. The computed radiography systems that complied with these regulatory requirements were the ones that showed themselves really proper for weld inspection. As a general conclusion of these requirements based on standards, the user must define the minimum SNR N required (see table 1) and SR b associated with the thickness of the material inspected, radiation energy and procedures of the service providers. Digital detectors should have the same or better SNR N than the film systems to compare. The minimum required SR b can be derived from the geometrical unsharpness (ug) requirements of standards as EN 444, EN 1435, ISO 5579, E 1032 and E1742. The recommended SR b value is about ½ of the ug requirement, which depends mostly from the material thickness [7]. The European Standard (5) and Petrobras show the SR b required for a radiography system based on radiation energy and thickness penetrated. This means that the double wire IQIs are critical in the computed radiography works, in the sense that they evaluate the spatial resolution of the system, and IQI of contrast (wire or hole) shall also be used to assure the acquisition of a proper contrast in the image. INAC 2007, Santos, SP, Brazil. 8

For the in-service inspection of wall loss, the quality requirements are not too strict, and there are no international standards about this theme; in Brazil, it seems that only Petrobras developed a specific standard to be released still this year. In this standard, the requirements refer to acceptable geometrical unsharpness and the saturation in grey levels at the end of the plates, where the intensity shall not exceed 90% of the limit of the dynamic range of the scanner. For example, for 12 bits (4096 grey levels), the limit would be 3686 levels. With regard to SNR N, it was noticed that the advisable minimum value of 50 would produce images with a good quality. 4. MINIMUM REQUIREMENTS FOR A COMPUTED RADIOGRAPHY SYSTEM This topic does not aim at addressing techniques used for in-service inspection of wall loss or weld inspection. It aims, based on all previous discussion, at establishing the minimum requirements for a radiography system in accordance with the application to which it is directed. There can be found in market scanners with laser spot sizes from 210 to 8 µm diameter. For the inspection of wall loss, this range services it well, however, for the weld inspection, just values inferior to 12.5 µm reach the quality required, for all situations of diameters and thickness [7, 8, 12]. As the laser spot size increases, there shall be restrictions regarding compliance with the requirements, with regard to limitations in the diameter range and thickness able to be inspected. This relationship between specification of equipment and limitations in radiographic inspection is not quite well defined in literature and experience of lab oratories and field, being a risk generating factor to the quality of inspection and probability of detection of defects. In relation to pixel size, for the inspection of wall loss, and with the use of the ordinary sources, it can be worked as from the maximum value of 200 µm (or 130 dpi). The experience shared between Petrobras and the European companies showed that in the inspection of wall loss there has been obtained very good results (probability of detection superior to 95%) with a 100 µm pixel size and laser spot size, what could show these values as advisable. For the weld inspection, the size ranges from 200 to 30 µm, according to the thickness of the material to be radiographed and energy or activity of the radiation source [5]. With the use of the High Resolution Computed Radiography Systems, isotopes with diameter larger than 1 mm should not be used for weld inspection, under penalty of affecting the quality of the results, except if for penetrated thickness superior to 40 mm. Finally, the dynamic range of the equipment must also be noticed. The ideal is that it is always equal or superior to 12 bits; for weld inspection, it must always be superior to this value. As high quality plates, the blue plates are outstanding for inspecting welds [7, 12]. It must be emphasized that the largest the plates are, there is a limitation of storage of data by the program and the scanning time is affected (the largest the plate, the longer is the time spent for a highly accurate scanning). There are no restrictions for corrosion monitoring and, of course, the quality of the plates shall interfere with the quality of the results. For the inspection of wall loss, there are no restrictions and, of course, the quality of the plates shall interfere with the quality of the results. It must be emphasized, however, that with the use of the High Resolution Computed Radiography Systems (scanner and high quality imaging plate), the inspection of the weld can be performed, but INAC 2007, Santos, SP, Brazil. 9

with similar exposure term or even an inspection term superior to the one spent with the conventional radiography [7, 8, 12]. In face of the limitations that still exist, the majority of the countries of the European Community still do not use the computed radiography in the weld inspection [12]. This is due to the small number of models of scanner that comply with the conditions required. 5. LONG TERM - STABILITY Another important issue and quite repeated [5, 13] refers to the long term stability of the CR systems. These standards show that the tests must be performed from time to time to evaluate the performance of the system along its useful life, so that the results can validate its quality or even refuse it. The frequency depends on the frequency of use and can be negotiated between the parties: contracting party and service provider. Petrobras standards recommend these tests are carried out annually, further requesting, at the same time, the evaluation of the luminance level of the monitor used for the reports. The quality parameters of the evaluated system are: modulation transfer function (MTF), basic spatial resolution, normalized signal-to-noise ratio (all three parameters evaluated in both direction of the plate), contrast sensitivity, dynamic range, fading, laser beam function, blooming, slipping and shading. To make an analysis of this parameter, it is recommended [5, 13] the use of a kind of IQI or Phantom, which is available in the market in the following dimensions: 14 x 17 and 8 x 10. The international standards [5,13] and Petrobras present a form that must be filed being the test reports reported periodically, and it shall operate as a certificate of system validation. Of course that these tests must be performed by an independent specialized lab. In Brazil, there is only one lab qualified to do this and Petrobras is encouraging the qualification of other units throughout the country. 6. CONCLUSIONS The Computed Radiography (CR) is a technique that presents a series of advantages known, but it must be used with discernment as there shall be much to advance in terms of quality of image obtained. The test procedures are still based on experiments, attempts and errors due to the lack of a consecrated methodology to choose the parameters, as it occurs in the conventional radiography. There are no international standards about the use of CR in the inspection of wall loss and, in Brazil, it seems that only Petrobras developed an internal standard about the subject. For the use in the inspection of wall loss, the scanners currently used comply with them as the requirements are not so severe. However, it is recommended the use of the minimum set-up as follows: 100 µm laser spot, maximum pixel size of 100 µm and dynamic range equal or higher than 12 bits. The mandatory appendix of ASME V, which addresses the CR, only makes requirements as to the control of the sensibility with the use of IQIs, what was shown insufficient to acquire radiographies of quality. INAC 2007, Santos, SP, Brazil. 10

The European Standards EN-14784, parts 1 and 2, as well as the US Standard ASTM-2446 provide requirements as to the values of normalized signal to noise ratio (SNR N ) and basic spatial resolution (SR b ), besides the use of IQIs. The computed radiography systems that complied with these regulatory requirements were the ones that showed themselves really proper for weld inspection. Few scanners available in the market present the specification required for the weld inspection. Thus, few scanners, even with the use of high quality plates, of the blue type, meet the requirements for weld inspection. The optimal specification for a scanner proper to weld inspection would be: laser spot size of 12.5µm, minimum available pixel size of 30-40µm and dynamic range higher than 12 bits. As the laser spot size increases, there shall be restrictions regarding compliance with the requirements, with regard to limitations in the diameter range and thickness able to be inspected. This relationship between specification of equipment and limitations in radiographic inspection is not quite well defined in literature and experience of lab oratories and field, being a risk generating factor to the quality of inspection and probability of detection of defects. The use of isotopes (Se75, Ir192 and Co60), jointly with high definition systems for weld inspection must be limited to source sizes of 1 mm diameter, at the most. Above this value, the quality of the image would be compromised. However, for thicknesses higher than 40mm, this harmful effect would be null. With the use of the High Resolution Computed Radiography Systems (scanner and phosphorous plate of high quality), the exposure time is similar or even the inspection term is superior to the one spent with the conventional radiography. In face of the existing limitations, the majority of the countries of the European Community still do not use the computed radiography to inspect weld threads. It is important that the service providers perform a periodical evaluation of its computed radiography system from time to time through a qualified and independent laboratory, in order to validate its use in the field. It is equally important that this evaluation is required by the persons that are contracting the services. REFERENCES 1. von Seggern, H.; Photostimulable X-Ray Storage Phosphors: a Review of Present Understanding, Brazilian Journal of Physics, vol. 29, no. 2, June, 1999; 2. Takahashi, K.; Progress in Science and Technology on Photostimulable BaFX:Eu 2+ (X= Cl, Br, I) and Imaging Plates, Journal of Luminescence, vol (100), pp 307-315, 2002; 3. Seggern, von H.; Photostimulable X-Ray Storage Phosphors: A Review of Presents Understanding, Brazilian Journal of Physics, 29, 1999; 4. Hangleiter T, et al; Temperature Dependence of the Photostimulated Luminescence of X- Irradiate BaFBr:Eu 2+, Journal of Physics. Condensed. Matter - 2, 6837-6846, 1990; INAC 2007, Santos, SP, Brazil. 11

5. BSI BS EN 14784 - Non-Destructive Testing Industrial Computed Radiography with Storage Phosphor Imaging Plates Part 1: Classifications of Systems, -2: General Principles for Testing of Metallic Materials Using X-Rays and Gamma Rays, 2005; 6. Munekawa, S., Ferrara, J.; Imaging Plate Detector for X-Ray Diffraction, disponível em http://bio.demokritos.gr/ccm/imageplates.html, July, 2005; 7. Ewert, U., Zscherpel, U., Bavendiek, K.; Replacement of Film Radiography by Digital Techniques and Enhancement of Image Quality, NDE 2005, National Seminar of Indian Society for NDT Proceedings, Decembre 2 4, Kalkutta, India, 2005; 8. Zscherpel, U., Ewert, U.; Computed Radiography, IAEA Training Course on DIR, January, 2006; 9. Lopes, R., Correa, S., Notas de aula do curso de Radiografia Industrial Computadorizada, Abende, São Paulo, May 2006; 10. ASTM 2446 - Standard Practice for Classification of Computed Radiology Systems, 2005; 11. Oliveira, F., Estudos de Sistemas de Radiografia Computadorizada com Uso de Placas de Fósforo, M.Sc., COPPE/UFRJ, March, 2004; 12. Dr. Uwe Zscherpel, from BAM (Bundesanstalt für Materialforschung und Prüfung), personal communication, Rio de Janeiro, March 2007; 13. ASTM 2445 - Standard Practice for Qualification and Long-Term Stability of Computed Radiology Systems, 2005. INAC 2007, Santos, SP, Brazil. 12