With the passage of two decades since its introduction, computed radiography (CR) has become a mainstream technology for acquiring ordinary radiographic projections in digital form. CR is extremely flexible, promising to replace virtually all conventional screen-film radiography, including mammography, scoliosis, and panorex studies. CR is especially suited to bedside radiographic examinations, traditionally the worst images in the hospital because of the exigent circumstances under which the examinations are performed. CR's wide latitude and automatic density adjustment dramatically improve the consistency of bedside radiographs. The appearance of the CR image can be modified as desired after acquisition. The CR image can be distributed virtually anywhere electronically, viewed simultaneously by multiple care providers, and reprinted, as necessary. The CR image is acquired on reusable image media (RIM), which can be erased and imaged thousands of times, eliminating the need for film and chemicals.

QC STILL CRITICAL

The success of CR leads to the misconception that quality control (QC) processes are no longer necessary. Early adopters of CR claimed that retake rates decreased to zero. In actual fact, QC processes for CR are no less important than they are for conventional screen-film radiography, and must be modified to take into consideration the unique characteristics of CR technology.

Fallacy No.1. No more bad images with CR. Traditionally, the primary cause of rejected radiographic images is mispositioning. Automation has not been invented to correct for mispositioning, patient motion, inadequate inspiration, wrong examination performed, wrong patient examined, improper collimation, double exposure, or the many potential artifacts that can interfere with the proper acquisition of a radiographic projection. It would be naive to believe that all these classic causes of rejected images would suddenly cease with CR. Indeed, analysis demonstrates that rejected images occur about as frequently in CR institutions from roughly the same causes as in conventional departments. Unrealistically low reject rates are often the result of inattention to the erasure of substandard electronic images (see Figure 1).

Figure 1. Image resulting from double-exposure of cassette. This error occurs in conventional screen-film radiography and CR, illustrating Fallacy No. 1.

Fallacy No. 2. CR is "plug-and-play." QC begins with appropriate acceptance testing of each CR system before it is introduced into clinical service. During installation, it is important to verify that all hardware, software, and durable goods that were purchased have arrived in good condition. Every component needs to be properly configured and integrated into the hospital's imaging operation, and testing should be performed to assure this. Every CR vendor has adjustment and calibration procedures specified in their service manuals. At an absolute minimum, these calibrations should be performed, verified, and documented. The manner in which patient demographic information is associated with the CR image must be configured and tested. The infinite variability with which the CR image can be acquired and modified before transmission and display allows a variety of errors in configuration by human operators (see Figure 2).

TRAINING IS KEY

Knowledgeable, skilled technologists must perform QC processes. Most technologist training programs do not address CR, and most technologists have no outside experience with CR. This implies that technologists must be trained about CR technology, initially by vendor applications personnel. Additional training on local practice, policies, and procedures must be developed and conducted by local radiology personnel. Customization of CR examination selections and the appearance of CR images is also a joint vendor-hospital responsibility.

Figure 2. CR image processed by parameter settings from two different CR scanners in the same facility. This configuration management problem is unique to CR.

Fallacy No 3. Use the same radiographic technique for CR that you used for screen-film.There is no doubt that a screen-film technique guide can be used to produce CR images. However, it is unreasonable to expect that a technique guide that was developed for a specific screen-film system would be optimal for CR, especially considering the multitude of speed classes for commercial screen-films. CR can be operated at many different speed classes. Objective data suggest that in order to match the noise characteristics of a typical screen-film system, CR must be operated at approximately 200 speed, corresponding to a medium speed screen-film. As for kilovolt peak (kVp), CR is reputedly less sensitive to kVp than most screen-films; however, CR's absorption "k-edge" is different from most screen-films in commercial use. A familiar conventional technique guide might be a reasonable place to begin.

Fallacy No. 4. You cannot cone down with CR. Technologists learn in their initial training that performing examinations with the collimator wide open causes more dose to the patient, degrades the image contrast, and produces more scattered radiation around the x-ray room. Keeping the collimator wide open is clearly bad practice in the case of CR, as well. Much of the software development in CR centered on detection of the collimator boundaries in images, so that contrast can be maximized for the region of interest within. Sometimes, especially with un-square, off-centered collimation or when the projection of one or more boundaries is interrupted by a solid object such as a pin, or with multiple fields projected on a single CR plate, the software can fail to find the boundaries. This can result in bizarre or nondiagnostic images. These images can usually be recovered by image processing without needing to repeat the exposure.

Fallacy No. 5. Our CR machines are calibrated at the factory. Although the CR machines may be calibrated at the end of the assembly line, much has happened between then and now. CR systems are complicated electro-optical mechanical systems. There is no guarantee that vibration and aging during shipment have not affected alignment and gain of the system. The manufacturer's x-ray generator may be quite different from that at the hospital. Actual field experience shows that the performance of CRs varies substantially from their specifications and that field calibrations drastically reduce variability between individual machines.

Fallacy No 6. That (exposure indicator) does not mean anything. One of CR's attributes is the ability to tolerate over- and under-exposure and still produce an image with consistent appearance. Unfortunately, CR's adjustable density means that density is no longer a reliable indicator of exposure factor control. Each major CR manufacturer (four at this writing) has developed a numerical indicator of the amount of radiation reaching the imaging plate. The absolute meaning of the indicator is subject to interpretation and technical interferences. For example, interpretation of the indicator depends on proper calibration of the gain of the CR. These indicators have been demonstrated to provide reliable measures of exposure to the plate for a given examination or view, but are more difficult to relate from one type of examination to another. Software failures, such as misidentification of collimation boundaries, can also yield spurious results. Nevertheless, exposure targets and ranges have been developed using the numerical exposure indicators and, for the most part, work well for monitoring exposure factor control in actual clinical practice.

Fallacy No. 7. Use whatever radiographic technique you please with CR. An even more dangerous fallacy than No. 3, this notion arises from the observation that the process of photo-stimulable luminescence on which CR is based produces a signal that is linearly related to the amount of radiation absorbed by the plate over at least a factor of 10,000! This ignores other important facts. First, any electro-optical system, such as CR, has a finite dynamic range, perhaps 1,000, and a limit of adjustment. Second, the radiographic projection of real clinical features has a range of exposures, perhaps a factor of 100 from most to least dense. In order to capture all the clinical features in a single exposure, the total tolerance for misexposure is a factor of 10. Underexposure by a factor of five or likewise overexposure by a factor of five risks losing clinical features. This description is not strictly true for CR systems with automatic speed adjustment: they probably have a tolerance factor of about 50. Not to worry: we have more than a century of experience controlling the amount of radiation produced by an x-ray generator, because screen-film is much less tolerant.

EXPOSURE CREEP CAVEAT

The next important fact is that the amount of noise in the image depends on the amount of radiation used to produce the image. A CR system operated at the limit of adjustment produces extremely noisy images. The more noise in the image, the less image processing that can be used without producing artifacts. Also, noisier images are more likely to be rejected by radiologists, who have been trained to be skeptical of noisy images, which may hide low contrast details of clinical importance.

On the other hand, using a lot of radiation produces CR images that are extremely crisp and unlikely to be rejected by a radiologist. Technologists are quick to notice this, resulting in a phenomenon known as "exposure creep." Using more radiation than necessary to produce a diagnostic quality image is clearly a violation of the "as low as reasonably achievable" (ALARA) principle and is a disservice to the patient.

Fallacy #8. CR image processing is absolutely optimized for the visualization of clinical features.

About the only universal agreement that you can find among CR practitioners is that no one likes the image processing that was delivered with their particular system. CR image processing was initially designed to make an unprocessed CR image mimic the appearance of conventional screen-film when printed by a laser camera. This approach is reasonable when you consider the worldwide acceptance of screen-film radiographs for displaying clinical features. Underlying assumptions are that screen-film rendering is the best that can be done, that the image is viewed in a static, not dynamic fashion, and that similar radiographic technique will be used to produce the images. As CR has been fielded, image processing has evolved. There is no guarantee that image processing that was optimal for viewing a transilluminated laser print of a CR image will also be appropriate for viewing a CR image displayed on a cathode ray tube (CRT) or flat panel.

Fallacy No. 9. There is no limit to the number of times a CR imaging plate can be used.

Sure enough, from a physics point of view, there is no limit to the number of times a CR imaging plate can be exposed to radiation, have its stored signal harvested, be erased, and exposed again. Even when bombarded with high energy particles from a cyclotron, permanent damage to the CR plate is not evident. However, there are mechanical, chemical, and industrial reasons why plates do not last forever. The plates are subject to scratches, scuffs, cracks, and contamination with dust and dirt, which can interfere with the image. One of the primary components of the plate is iodine, which oxidizes to iodate on exposure to water. When this occurs, the plate becomes discolored and the result is artifacts in the image. Over the years, new formulations have been devised for imaging plates to enable or exploit the features of new CR scanner mechanisms. Unfortunately, the modified formulation may be required by the new mechanism, making the older plates obsolete. Likewise, the new formulation may not be backward compatible, making the older mechanism obsolete.

Fallacy No. 10. Any errors in a CR image can be corrected by image processing.

CR vendors have made dramatic progress in the features and functions of computer workstations that can be used for quality control before releasing the images to distribution and archive. These workstations are especially useful for correcting patient demographic and examination information as well as verifying proper exposure indicator values. In contrast to older systems, most commercially available CRs today retain raw image data to accommodate reprocessing. However, the discussion of No. 1 above provides a litany of errors in CR images that cannot be corrected by image processing. Image processing cannot improve the visibility of clinical features that are just not present in the raw image, because they are outside the radiation field or off the edge of the imaging plate, or because the examination was performed with such an inappropriate technique that there was no contrast between the clinical feature and the surrounding tissue. Image processing is a poor substitute for proper examination technique.

CONCLUSIONS

The important lesson to be learned is that CR is not a panacea. CR is just another tool for doing the business of radiology. We can choose to do this business in a professional manner or in a less efficient, haphazard fashion. Errors can still arise in the practice of CR, and QC processes must be in place to detect and correct those errors when they occur. These QC processes are strikingly similar to the processes we used routinely in conventional screen-film radiology.

BIBLIOGRAPHY

• Freedman M, Pe E, Mun SK, Lo SCB,  Nelson M. The potential for unnecessary patient exposure from the use of storage phosphor imaging systems. Proceedings of the International Society for Optical Engineering. 1993;1897:472-479.
• Gur D, Fuhman CR, Feist JH, Slifko R, Peace B. Natural migration to a higher dose in CR imaging. Eighth European Congress of Radiology; September 12-17, 1993; Vienna. Abstract 154.
• Honea R, Blado ME, Ma Y. Is reject analysis necessary after converting to computed radiography? J Digit Imaging. 2002;15(Suppl 1):41-52.
• Huda W, Arreola M, Jing Z. Computed radiography acceptance testing. Proceedings of the International Society for Optical Engineering (SPIE). 1995;2432:512-521.
• Huda W, Slone RM, Belden CJ, Williams JL, Cumming WA, Palmer CK. Mottle on computed radiographs of the chest in pediatric patients. Radiology. 1996;199:249-252.
• Oestmann JW, Prokop M, Schaefer CM, Galanski M. Hardware and software artifacts in storage phosphor radiography. Radiographics. 1991;11:795-805.
• Samei E, Seibert JA, Willis CE, Flynn MJ, Mah E, Junck KL. Performance evaluation of computed radiography systems. Med Phys. 2001;28:361-371.
• Seibert JA. Photostimulable phosphor system acceptance testing. In: Seibert JA, Barnes GT, Gould RG, eds. Specification, Acceptance Testing and Quality Control of Diagnostic X-ray Imaging Equipment. Woodbury, NY: American Association of Physicists in Medicine; 1994:771-800. Monograph No. 20.
• Solomon SL, Jost RG, Glazer HS, Sagel SS, Anderson DJ, Molina PL. Artifacts in computed radiography. AJR Am J Roentgenol. 1991;157:181-185.
• Volpe JP, Storto ML, Andriole KP, Famsu G. Artifacts in chest radiography with a third generation computed radiography system. AJR Am J Roentgenol. 1996;166:653-657.
• Willis CE, Weiser JC, Leckie RG, Romlein J, Norton G. Optimization and quality control of computed radiography. Proceedings of the International Society for Optical Engineering (SPIE).1994;2164:178-185.
• Willis CE, Leckie RG, Carter J, Williamson MP, Scotti SD, Norton G. Objective measures of quality assurance in a computed radiography-based radiology department. Proceedings of the International Society for Optical Engineering (SPIE). 1995; 2432:588-599.
• Willis CE, Mercier J, Patel M. Modification of conventional quality assurance procedures to accommodate computed radiography. S/CAR '96 Computer Applications To Assist Radiology. Great Falls, Va: Society for Computer Applications in Radiology;  1996:275-281.
• Willis CE. Computed radiography imaging and artifacts. In: Siegel EL, Kolodner RM, eds. Filmless Radiology. New York: Springer-Verlag; 1999:137-154.
• Willis CE. Computed radiography: QA/QC. In: Practical Digital Imaging and PACS. Medical Physics Monograph No. 28. Madison, Wis: Medical Physics Publishing; 1999b:157-175.
• Willis CE. Quality control of computed radiography systems. Quality assurance for the third millennium. American Association of Physicists in Medicine Southwest Regional Chapter Spring Meeting and Conference; May 2, 2000; Puerta Vallarta, Mexico.
• Willis CE. Quality assurance: an overview of quality assurance and quality control in the digital imaging department. In: Quality Assurance: Meeting the Challenge in the Digital Medical Enterprise. Great Falls, Va: Society for Computer Applications in Radiology; 2002:1-8.

Charles E. Willis, PhD, is a medical physicist at Texas Childrens Hospital, Houston, and an associate professor at Baylor College of Medicine.