Digital Radiography in Industry: Digital Detector Arrays in Radiographic Testing by Brad Kraai

Vol. 15,13, No. 31 FOCUS Digital Radiography in Industry: Digital Detector Arrays in Radiographic Testing by Brad Kraai

Introduction This article explores DDA systems: application, capability, process controls, Digital detector array (DDA) systems image attributes and evaluation, and within industrial radiography are becoming personnel qualifications—to hopefully very common in high performance, critical promote an increased interest and margin of safety, and endurance test enlightenment for potential users. article inspections. Investment castings, thermal joints, and a wide variety of other Definitions test articles are being routinely inspected, Per ASTM E 2736, Standard Guide for with improved probability of detection Digital Detector Array Radiology, a digital (POD), and much faster throughput detector array is defined as: “an electronic than conventional film systems—with device that converts ionizing or penetrating high levels of user satisfaction. Potential radiation into a discrete array of analog users within the industrial X-ray imaging signals which are subsequently digitized community continue to recognize these and transferred to a computer for display values and exploit this new technology as a digital image corresponding to the for potential applications, but to do so, radiation energy pattern imparted upon the a basic understanding of DDA systems input region of the device. The conversion and application is a necessary foundation. of the ionizing or penetrating radiation Smooth implementation of this novel into an electronic signal may transpire by technology can be challenging, and first converting the ionizing or penetrating consultation should be considered from an radiation into visible light through the use outside, unbiased, reputable organization. of a scintillating material” (ASTM, 2010a). There are several integrators, or vendors, While DDAs can be used for real-time or that are widely recognized as providers of radioscopic techniques, most applications DDA systems development, installation, for critical test articles employ static and service—each having its own merits imaging and evaluation. Figure 1 provides and specialties. a simplified diagram of a DDA.

the responsibility of the potential user, not the integrator or vendor, to ensure that DDA system performance metrics are met for a given target application. X-ray signal Obviously, DDA systems are a considerable capital expenditure, so once the system is designed, built, and installed, it must Test article meet the requirements—resolve the defects reliably—as intended. This cannot be overemphasized. Film images are commonly used as the baseline or referee for a correlation study with the DDA imaging techniques, and will often be Electronics shielding (reduces radiation dose to electronics) required for process approval. Gadolinium oxysulfide or cesium iodide (converts X-ray signal to light signal) High levels of automation are possible Digital data to Amorphous silicon TFT (converts light signal to electronic signals) frame buffer and often used with DDAs. Imaging, from Electronics (captures electronic signals, organizes and converts to digital data) one acquisition to the next, can be mere

Glass substrate (stable TFT and electronics structure base) seconds, depending on the type of DDA and acquisition settings, technique, and Figure 1. Simplified diagram of a digital detector array (not to scale). TFT = thin film transistor. level of automation. Fully articulating and programmable robotics within custom- built radiation enclosures are becoming The most popular types of DDAs in use raw or full fidelity image is saved unaltered quite common in turbine blade inspection; contain an initial and indirect conversion for critical applications. Raw image file size a single blade can be imaged with several layer (or scintillator)—typically either primarily depends on the resolution, or views in less than a minute. Other gadolinium oxysulfide terbium doped, pixel pitch, and input region (or array size) automation tactics may include external or cesium iodide thallium doped—that of the DDA, as well as its bit depth. ingress of test articles into the enclosure by converts an X-ray signal to visible light As one can see, DDAs are quite conveyors, or manual loading of numerous or luminescence. This luminescence then sophisticated electronic conversion parts in fixtures or platens. DDA systems enters the amorphous silicon (αSi) thin devices. Yet, they can provide simplicity with turntables for the test article and a film transistor diode array (discrete pixel to the radiographic imaging process, as C-arm—with the X-ray tube on one end locations), whereby the light is converted to consistency and reliability are often the and DDA on the other—are commonly an electronic voltage (bias change) at each result of a properly engineered, tuned, and used for larger test articles. Other pixel, which is subsequently “read out” of applied DDA system. sophisticated systems have incorporated the array in channels and groups during Applications two robots: one for the X-ray tube and the X-ray exposure. This electronic information other for the DDA, while the test article is amplified and then digitized, typically For the purpose of application validation, a remains stationary. System integrators can through several analog-to-digital converters, potential user must evaluate “representative be creative, and depending on the target synchronized, and sent to the frame buffer quality indicators,” or, more simply application, will often readily collaborate within the image processor and system stated, potential test articles with known with the potential user to design and software. conditions or defects (of a minimum size incorporate optimized article handling and Gray values for each pixel’s digital and all types likely to be encountered) for imaging for the intended application. (binary) value are then assigned by a lookup any DDA system under consideration. It must be realized that DDAs are in table, and a corresponding pixel matrix Quite often, DDA systems are designed fact a consumable, and will eventually that represents X-ray attenuation from the and built around a target application or require replacement due to performance initial image acquisition is generated and family: the potential test articles to be issues or degradation. They will and do considered as the raw or full fidelity image inspected with the DDA system will all fail (unpredictably, in most cases) over file. This image file also contains meta-data, be related—similar in material type and time. The cumulative effects of radiation or image tags of process information as material thickness, subject contrast range, exposure and thermal variations within configured by the system integrator. The and inspection standard criteria. It is the device produce effects that may range

2 · Vol. 15, No. 3 from image quality degradation all the resolutions (discrete pixel dimensions recommended 3 × 3 pixel matrix in the way to failure. ere are no guarantees or “pixel pitch”) currently available, the IQI hole decreases in turn. Higher M of a DDA’s usable duration. Inadvertent contrast sensitivity for these devices is factors increase pixel density for a speci c or unintentional exposure to the DDA remarkable—provided noise is controlled. dimension within the image, thereby should be avoided. Recent developments in Due to inherent DDA resolution, promoting higher POD. Higher M factors hardened electronics have increased DDA geometric magni cation (M) techniques applied within techniques will decrease the resistance to exposure e ects. Shielding are often employed, which amplify image eld of view, resulting in decreased of the electronics around the periphery discontinuity size and enhance POD throughput for larger test articles as of the DDA must be provided by the via a higher number of pixels under the compared to the DDA input region. integrator, and should be analyzed by the potential discontinuity or feature within Another primary consideration for any DDA manufacturer for warranty viability. the test article. It is a relatively simple geometrically magni ed technique is image

Moreover, users of DDA systems must matter to calculate for a speci c number unsharpness (Uimg), wherein a reduced contemplate detector failure or sub-par of pixels under a known dimension at e ective focal spot size (EFSS) is often imaging contingency and warranty when geometric magni cation, and quite often, necessary at higher magni cations. Uimg negotiating procurement. the calculation is applied to determine calculations take into account the

As mentioned previously, the DDA pixel density within an essential image geometric unsharpness (Ug) of the and its supportive software are very quality indicator (IQI) hole dimension. technique, (where [M – 1] × EFSS = Ug), complex. is DDA system complexity e minimum recommendation is three the geometric magni cation factor can be very intimidating, so a word of pixels, so by using the basic spatial (M = source-to-detector distance / source- advice: complexity of the entire DDA resolution (SRb) or e ective pixel size of to-object distance), and the SRb of the system should be held to a minimum— the DDA, and the speci ed IQI hole size DDA. Uimg can also be evaluated by con gurations should be limited to t the (d), minimum geometric magni cation can imaging, at the source side of the area needs of the user. During DDA system be calculated by the following equation of interest, a unique IQI—the duplex design and development, it is vital that (Figure 2). wire gage (Figure 3)—to determine the potential user recognize excessive which speci c wire pairs merge within complexity for what it can be—and often SR the acquired image: U then being (1) M =×4.25 b img is—the enemy of execution. Careful min d regarded as that speci c wire diameter and analysis of all DDA system features and its adjacent space combined. U and For example, if SR = 215 µm img their intricacy may reveal unwanted or b geometric magni cation are both important (0.008 in.) and d = 508 µm (0.020 in.), unwarranted complication. technique parameters that must be well M = 1.7987X (1.8X). Obviously, as min understood and evaluated prior to DDA the SR value decreases (for higher Capabilities b system build. resolution DDAs having a ner pixel While DDAs are challenged for image Several techniques exist for minimizing pitch), the M required to achieve the spatial resolution, due to the nite min noise: acquired image noise can never

SRb (effective IQI hole ø – d pixel size)

Figure 2. A 3 × 3 pixel matrix in image quality indicator (IQI) hole—calculation for minimum Figure 3. Duplex wire gage. magniﬁ cation.

TNT · July 2016 · 3 FOCUS | Digital Radiography in Industry be fully eliminated, and is composed of the law of Poisson exists, where noise will scale of film—that is, 1.50 to 4.00 H&D signal and system noise. Common noise increase by the square root of the variable (Hurter and Driffield), or ~250 usable reduction tactics include frame averaging, signal increase (exposure). That being said, radiographic optical densities. The usable DDA calibration, X-ray beam filtration, higher exposure or signal provides higher range of the DDA will always be less X-ray beam collimation, and exposure SNR, albeit with an increase in exposure than the full scale or bit depth, but is still optimization. The averaging of frames time or image acquisition, and effectually significant. This long scale enables the basically provides better statistics for a higher dose to the DDA. High SNR very low contrast sensitivity mentioned the raw image data, by adding the pixel results in extremely low contrast sensitivity previously—provided noise is controlled. signal value for each individual pixel capability for DDA techniques: quite The long scale also provides adequate image through all frames captured, dividing by often well below 0.5%, particularly for quality through a wider thickness range in the number of frames, and assigning the thinner areas of interest within test articles. one exposure. Another advantage of DDAs averaged value to that pixel for the raw This high SNR also contributes to high is this very high dynamic range as compared image. Averaging more frames impacts contrast-to-noise ratio (CNR), as measured to short scale imagers, mainly film. throughput, increases dose to the DDA, within an image of a conventional IQI. As can be clearly seen, there are and typically only presents improved Figure 4 illustrates a CNR measurement multiple variables that must be evaluated statistics at a threshold. DDA calibrations, on a 0.05 IQI, 127 µm (0.005 in.) T, on a and correlated during DDA system or normalizing, are a necessary function 1.27 mm (0.05 in.) thick base material. As consideration and design. Resolution and of the DDA imaging process. Most users important technique attributes, SNR and contrast performance for any radiographic apply calibration techniques for DDA offset CNR measurement can be accomplished imaging system are the primary objectives, and gain per the integrator or vendor’s with specific tools within most imaging and the requirements for the inspection recommendation. Offset is basically dark software. High CNR values are not application must be recognized and or inherent noise within the DDA, and possible without high SNR values, and understood. The characterizations of gain is inconsistencies in pixel responses these two metrics can be used to validate performance within any DDA system are (amplitude or gain) during exposure. one another—provided measurements are various, but can be boiled down to system

Without normalization, any DDA would consistent and controlled. SRb, allowable or required geometric basically be unusable for an application. magnification, Uimg, and contrast sensitivity It is extremely important that the user through the material type and thickness establish DDA normalization protocol and range to be inspected. Many other evaluate its efficacy. X-ray beam filtration considerations exist, more or less depending is often employed, thereby attenuating on the target application, that are beyond certain portions of the soft radiation the scope of this article. Realistically, at the spectrum, which in turn reduces noise beginning, middle, and end of the day, the and improves signal-to-noise ratio (SNR). required performance of the DDA system Experimentation is key, and SNR can be is just this: to consistently and reliably evaluated during system validation and detect and accurately portray substandard technique development. Applying filtration defects for the selected application and its also reduces dose to the DDA, which technique. may increase its lifespan. Collimation, or restricting the X-ray beam to only Process Controls illuminate the input region of the DDA, Any DDA system’s performance must be will also lower signal noise and reduce dose monitored for instability and degradation. Contrast-to-noise ratio to the DDA’s electronics. Besides fixed Figure 4. Most users apply a duplex plate phantom measurement on an image quality collimators, there are also programmable indicator of 29.5; this translates to (DPP) and specialized software to enable X-ray beam shuttering mechanisms that 0.8% contrast sensitivity. consistency within the measurements can be completely closed (for X-ray system required for stability and performance. warm-up) or opened/closed, as appropriate, The DPP (Figure 5) must be of the for the imaging technique. Proper exposure The available DDAs are normally 14-bit same material group as the inspection ensures optimized signal capture within (214) or 16-bit (216), 0 to 16 383 or 0 to application, with a thin and thick section the DDA, and a translation to low contrast 65 535 possible pixel values, respectively: closely matching the thinnest and sensitivity. Within any signal amplification, very long scale as compared to the limited thickest sections of the test article(s).

4 · Vol. 15, No. 3 qualify all acquired images to the previous qualifier. If the qualifier does not present the required level of image quality, all previous images to the last viable qualifier are then considered invalid. The user must assess the reprocessing risk and possible burden associated with qualifier frequency and application. Not unlike any nondestructive testing process, practices for process controls must demonstrate system repeatability and reliability, providing confidence and value for the user and customer. Figure 5. Duplex plate phantom—used for digital detector array system stability measurements. Ti6Al4V titanium alloy 19 to 1.27 mm (0.75 to 0.05 in.) including duplex Image Attributes and Evaluation wire gage. Any digital image is a matrix of pixels, each pixel having a relative value and X, Y coordinate address (where X = column ASTM E 2737, Standard Practice for Digital uncorrectable bad pixels—referred to position and Y = row position)—pixels Detector Array Performance Evaluation as cluster kernel pixels (CKPs)—may are also defined as the smallest component and Long-term Stability is typically increase the risk of missed defects to the of the image that may be altered. The requisite as a process control mechanism point that DDA replacement is necessary. acquired DDA image, being of very (ASTM, 2010b). The performance Avoiding imaging within the area of long scale, requires image processing or attributes measured can include: spatial the DDA that contains CKPs is also an alteration across the matrix, to render resolution, contrast sensitivity, SNR, acceptable practice but normally involves discontinuity or feature visibility with signal levels, offset, burn-in, lag, and bad reprogramming for test article positioning human vision. There are many different pixel distribution. Most users apply the and possible reductions in usable field processing techniques than can be applied: core image quality tests, and by using of view. DDA bad pixel and CKP spatial filters, window/level adjustments, statistical process controls, upper and lower considerations are all about management, and electronic magnification (zoom) being thresholds of ±3σ are determined during through mapping the DDA and evaluating most often utilized. Any image processing a prescribed baseline period. Bad pixel the bad pixel distribution list, and most techniques must be validated, documented, distribution includes mapping and listing for importantly the presence of CKPs, at and controlled—it could be said that there pixels that are outside of a specific standard, predetermined intervals or as necessary. The is a viewing technique as well as an imaging as defined within ASTM E 2597, Standard DDA system user or agent should evaluate technique due to the variety of tools at the Practice for Manufacturing Characteristics of the bad pixel map and distribution list of technician’s disposal. Image discontinuities, Digital Detector Arrays (ASTM, 2014). All the purchased DDA upon system buyoff features, and attributes can be analyzed DDAs have bad pixels, and DDAs may be and also evaluate the DDA for bad pixels or quantified with various tools contained graded according to the percentage of bad upon system delivery. within most software platforms. Line or pixels within the device upon manufacture. Ongoing production image quality is area profiles, linear measurements, regions Connected bad pixels or clusters, bad lines most often verified by the use of a qualifier of interest for SNR/CNR measurements or segments (columns or rows), and their image, or image capture of a specific or local image processing, histograms, presence within a DDA will increase over quality indicator (conventional IQIs or and other implements are available for time. Most of these pixels in the DDA are other) at a prescribed frequency, due to these purposes. Annotation tools are correctable by interpolation within the the automated inspection environment. used to identify and characterize image DDA software; those bad pixels without It is not necessary, nor practical in most indications, and can either be saved as enough good neighbors (five minimum cases, to include an IQI within each image an overlay, or placed permanently within are required per ASTM E 2597) cannot acquisition. Qualifier images are acquired the image provided they do not mask or be corrected and must remain in the and evaluated at a prescribed frequency, obscure an area of interest. Training and production image, and are therefore usually at a minimum of once per shift as experience, and scripted work instructions, considered image artifacts. Depending determined by the user and approved by complement the accuracy and consistency on the application, an escalation of these the end user or customer, and these images of image evaluation activities.

TNT · July 2016 · 5 FOCUS | Digital Radiography in Industry

is the revision of the user’s written practice for qualification and certification of DDA personnel. Most users are making the transition from film-based radiography, so initially, the radiographic testing (RT) Level 3 must be considered for DDA Level 3. Formal training is required, and any experience gained toward qualification includes, but may not be limited to, DDA system design and application studies, vendor demonstrations, and technical conferences. Experience should be accumulated under supervision or guidance of an examiner, instructor, or outside Figure 7. Full automated digital detector Figure 6. Digital viewing environment. agency. RT Level 2 promotion to DDA array (DDA) system: 225 kVp µ-focus tube (left); 200 µm amorphous silicon DDA Level 2 begins with 40 h of formal training (center); fully articulating robot and gripper and the required hours of on-the-job (top); and “pigeon hole” casting carrier The digital image display should be a high training/experience (OJT/OJE). The formal (right). performance, medium to high resolution, training should be administered by an liquid crystal display (LCD) or LCD/light- experienced, requirement and application emitting diode type. Typically, medical savvy instructor, and should occur prior to, efficiency, investment returns, and user grade monochrome displays are incorporated or in conjunction with, OJT or OJE. The satisfaction that DDA systems are capable into the DDA system viewing workstation. OJT/OJE should be well organized and of providing. h High luminance (brightness), as measured in delivered under structured planning and candelas per square meter (cd/m2) at 100% documentation. Technicians can then be ACKNOWLEDGMENTS Figure 7 was provided by X-R-I Testing, Troy, digital driving level (full white) is warranted, fully qualified by examinations as required Michigan; and Pinnacle X-Ray Solutions, to promote high visible and quantified by the employer’s written practice and Suwanee, Georgia. contrast. A video test pattern, most often the subsequently certified. An audition period Society of Motion Picture and Television or secondary review of a newly certified REFERENCES Engineers’ SMPTE RP-133 or an approved DDA Level 2’s image evaluation is often ASTM, ASTM E 2597, Standard Practice appropriate, and may be required by the for Manufacturing Characteristics of Digital variant, is presented and evaluated for spatial Detector Arrays, ASTM International, West precision, visible contrast, and measured end user or customer. Conshohocken, Pennsylvania, 2014. Starting from scratch would require contrast performance of the display ASTM, ASTM E 2736, Standard Guide for (SMPTE, 1999). a similar regimen as film radiographer Digital Detector Array Radiology, ASTM The digital image viewing ambient qualification and certification for the DDA International, West Conshohocken, environment requirements are basically Level 2. Pennsylvania, 2010a. identical to those of radiographic film ASTM, ASTM E 2737, Standard Practice for Conclusion viewing, but the evaluation of soft copy Digital Detector Array Performance Evaluation and Long-term Stability, ASTM International, images has proven to be much more Increased radiographic inspection West Conshohocken, Pennsylvania, 2010. consistent, ergonomic and less rigorous throughput, higher POD and disposition accuracy, reduction in consumable costs, SMPTE, SMPTE RP-133, Specifications for than hard copy image handling and Medical Diagnostic Imaging Test Pattern for viewing, resulting in higher levels of and the anticipated solid return on Television Monitors and Hardcopy Recording technician confidence, performance, and investments for DDA system users can Cameras, Society of Motion Picture and contentment (Figure 6). be realized with proper understanding, Television Engineers, Inc., White Plains, New planning, and implementation (Figure 7). York, 1999. Personnel Qualification Potential users must decide if outside AUTHOR Potential users of DDA systems also assistance is necessary—at the beginning Brad Kraai: X-Ray Industries, Inc., an have the challenge of fully qualifying of the DDA system selection process— Applus RTD Co., NDT Training, Consulting technicians and Level 3s for this new instead of during or after. This proactive & Services Group; e-mail brad.kraai@ imaging process—the beginning of which approach can facilitate the process xrayindustries.com.

6 · Vol. 15, No. 3