- Email: [email protected]
Computed radiography X-ray exposure trends
Computed Radiography X-Ray Exposure Trends J. Anthony Seibert, PhD, David K. Shelton, MD, Elizabeth H. Moore, MD
Rationale and Objectives. Computed radiography provides correct optical density on film, independent of the incident radiation exposure, but it can result in under- or overexposure of the imaging plate. In the current study, we evaluated the radiation exposure trends of computed radiography over a 2-year period for portable chest examinations to determine and compare the radiographic techniques of the computed radiography system relative to conventional screen-film detectors. Methods. A Fuji computed radiography system was interfaced to a digital workstation to track system usage and examination demographics, including examination type and sensitivity number. Hard-copy films were used for diagnosis. The sensitivity number, a value inversely related to incident exposure on the imaging plate, was used to determine whether the proper techniques were used by the technologists. Results. The initial use of the computed radiography system revealed a broad distribution of exposures being used; complaints regarding noisy films (e.g,, underexposure) resulted in subsequent overexposure for a significant number of films. A quality-control audit indicating excessive exposure resulted in educational feedback and a tighter distribution of exposures within the optimal range as determined by our radiologists, The average technique was approximately equivalent to a 200-speed system.
Received September 18, 1995, and accepted for publication after revision November 29, 1995.
Conclusion. Computed radiography provides excellent dynamic range and rescaling capabilities for proper film optical density, and th~_ts fewer repeat examinations. However, underexposure results in suboptimal image quality that is related to excessive quantum mottle. Overexposure requires film audits to limit unnecessary radiation exposure. In general, the optimal e:kposures are achieved with approximately 1.5=2, times the incident detector exposure of a 400-speed rare-earth system. The ability of computed radiography to reduce radiation exposure is unlikely when compared with a typical rare-earth screen-film combination (400 speed) in terms of adequate image quality for the diagnosis of subtle, low-contrast findings. For certain diagnostic procedures (e.g., nasogastric tube placement verification), lower exposures can be tolerated.
Acad Radio/1996;3.'313-318 © 1996, Association of University Radiologists
Key Words. Computed radiography; radiation dose; portable chest imaging; photostimulable phosphor detectors; digital X-ray imaging.
From the Department of Radiology, University of California Davis Medical Center, Sacramento, CA. Parts of this article were presented at the 43rd Annual Meeting of the Association of University Radiologists held in San Diego, CA, in May 1995. Address reprint requests to J. A. Seibert, PhD, Radiology Research, University of California Davis Medical Center, FOLB-II-E, 2421 45th St., Sacramento, CA 95817.
313
SEIBERT
ET AL.
omputed radiography is the name given to the pro-
cess of digital image acquisition using an X-ray C detector comprised of a photostimulable phosphor imaging plate and an image reader-writer that processes the latent image information for subsequent brightness scab ing and laser printing on film [1]. Separation of the acquisition and the image display events using digital acquisition and processing of the signal permits the correct optical density range to be achieved in the final image, independent of the exposure to the imaging plate. This is attributed to a linear signal response over four orders of magnitude (i.e., 104 or 10,000 times) in incident exposure. An internal scaling algorithm within the computed radiography reader corrects for under- or overexposure caused by inappropriate radiographic techniques. This scaling algorithm also remaps the variations in exposure in a manner similar to a film characteristic curve response and provides a consistent optical density on the printed film. These are significant benefits of computed radiography compared with screen-film systems. On the other hand, limiting the spatial resolution of the computed radiography image is much less than screen-film, and the X-ray detection efficiency of the imaging plate is less than a corresponding 400-speed rare-earth screen. In addition, secondary noise sources caused by luminance variations during the readout process [2], as well as electronic signal variations during digitization, increase the perceptible image noise for a given incident exposure. With current technology, the images are most often presented with a reduced film format size, making comparisons to full-size films suboptimal. Despite these drawbacks, improvements in technology, decreasing system costs, and a tangible future for electronic imaging in medicine have spurred an interest in clinical implementation of such systems. With expanding use, computed radiography is expected to be the frontline technique for acquiring digital large-area projection images in the coming years. It is therefore important to understand the features and limitations of these systems for X-ray projection imaging. There are reports demonstrating the ability to reduce exposures with computed radiography compared with conventional screen-film imaging without losing diagnostic accuracy [3, 4]. Exposure reductions of up to 50% have been claimed, although the speed of the screenfilm detector was not specified [3]. Comparison with a 200-speed system for pediatric applications also indicated an ability to reduce exposures [5, 6]. Other researchers have reported an increase in exposures
314
VoJ. 3, No. 4, April 1996
required with-computed radiography relative to a 400or 600-speed conventional screen-film system [7, 8]. A common theme to all of the investigations is the "potential" to reduce exposure by computed radiography depending on the examination type and the diagnosis requested. On the basis of this background, the purpose of this study was to evaluate the radiation exposure trends of computed radiography over a 2year period for adt~llt portable chest examinations at the University of California Davis Medical Center (UCDMC) and to compare the results with a conventional 400speed screen-film detector. We selected adult portable chest examinations as the focus because of the large number of examinations and the large number of repeat examinations, with the majority being caused by under- or overexposure (i.e., films were too light or too dark for optimal diagnosis). Initially, we thought that computed radiography would be an uncompromising solution to many of the problems encountered with suboptimal radiographic technique encountered with screen-film detectors. Because under- or overexposure is not readily apparent on the resultant computed radiography film images, standardization of the radiographic technique to determine the exposure range that provides reasonably lowqmage noise at a minimum exposure level without compromising the quality of the image was the goal of our study. MATERIALS AND METHODS
A calibrated Fuji computed radiography system (model AC-1 plus; Fuji, Tokyo, Japan) using standard resolution (ST-III-N) imaging plates produced clinical portable chest images on laser-printed film. Over the 2-year period of the study, approximately 30% of all adult portable chest images were acquired with computed radiography. Typical parameters used by the technologists were 85 kVp and 5-10 mAs and no antiscatter grid. The processing algorithm for the portable chest image was modified to print a single 9 x 11 inch (23 x 28 cm) image on reduced-format film (a two-thirds reduction factor of a standard 14 x 17 inch [35 x 43 cm] film). Over an 18month period, from July 1993 through December 1994, computed radiography examinations (both image and demographic information) were automatically transferred to a computer workstation via a digital acquisition system interface. The digital (soft-copy) image was not used for viewing or diagnosis. Information was collected regarding exam type, sensitivity number, plate size, and image latitude for all computed radiography examina-
Vol. 3, No. 4, April 1996
COMPUTED
tions, and stored on a month by month basis. Adult portable chest examination information was sorted from all other studies using a personal computer spreadsheet program. The sensitivity number was used to estimate the average incident exposure on the imaging plate for each image as follows: incident exposure (mR) =
200 sensitivity number
where 1 mR = 2.58 x 10-7 C/kg air. Thus, a sensitivity number of 200 indicates an approximate incident exposure of 1 mR (2.58 x 10-7 C/kg) to the imaging plate, whereas a sensitivity number of 600 represents an incident exposure of approximately 0.3 mR (0.774 x 10 - 7 C/kg). Verification of the sensitivity number accuracy was accomplished during the acceptance testing and subsequent quality-control checks on the computed radiography unit using a defined quality-control methodology [9]. Any retake images were consolidated with the total number of computed radiography films evaluated, and these represented approximately 1% of the total, based on a retake rate analysis. (Conventional screen-film images were not included in this evaluation.) Sensitivity numbers above 20,000 were excluded from the data set as u n e x p o s e d imaging plates. Incident exposure estimates were plotted as a function of the number of examinations (i.e., a histogram). Quarterly reports were presented to a radiology quality improvement committee for analysis, review, and determination of exposure trends for portable chest imaging using computed radiography. An opportunity for educational feedback to the technologists was channeled through the pulmonary radiologists, who qualitatively rendered a judgment about the film quality regarding perceived image noise. This communication was used in conjunction with image audits of the computed radiography data files to determine the "optimal" image quality in terms of sensitivity numbers (incident exposure on the detector), as tracked over the evaluation period. These exposure limits, were then used retrospectively on the 18 months of acquired exposure data. RESULTS
The initial use of the computed radiography system (over the first year of evaluating sensitivity numbers) revealed that there was a broad distribution of exposures being used in portable chest examinations, as shown in Figure 1 for the second half of 1993. The incident exposures estimated from the sensitivity numbers
RADIOGRAPHY
X-RAY
EXPOSURE
Adult portable chest Second half, 1993
~--
#exams 600
4824 total exams
58.1%
30.1%
TRENDS
Target exposure range 11.8%
500
Q3-400 Q4 . m 300 200
,,o.
100 0
700 550 450 375 325 2
225 175 125 75 <50
Sensitivity number Low -
Incident Exposure
= High
FIGURE 1. The number of portable chest X-ray examinations plotted as a function of incident exposure to the computed radiography imaging plate as determined by the sensitivity number. Adult portable chest examinations for the second half of 1993 are consolidated by quarter (July to September = quarter 3 [Q3] and October to December = quarter 4 [Q4]). Each point on the c u r v e represents a range of sensitivity values shown by the x-axis labels (not all points are labeled). The total number of films and percentages within each exposure range are listed at the inside top of the plot.
Adult portable chest First
half,1994
Target exposure range
~--
#exams 600
4572 total exams
38.3%
53.9%
7.8%
500
400
QI m
300
Q2 d
20O
.ir/Q~,
IO0 0
./J~ .,e"*')
700 580 450 375 325 2:
~-,..z% 225 175 125 75 <50
Sensitivity number Low "=
Incident Exposure
=- High
FIGURE 2. Adult portable chest examinations for the first half of 1994 ~.re consolidated by quarter (January through March = quarter 1 [Q1] and April through June = quarter 2 [Q2]). The total number of films and percentages within each exposure range are listed at the inside top of the plot.
were uniform across a broad s p e c t r u m / w i t h 30.1% of the films falling in the low (sensitivity > 275) and 11.8% of films falling in the high (sensitivity < 125) exposure ranges. During the first quarter of 1994, the technologists were told of the exposure evaluation study and the presence of the large number of high-incident exposures (low sensitivity numbers). A trend toward lower exposures occurred in quarter 2 (Q2), as illustrated by the comparison of Q1 and Q2 curves in Figure 2. A decrease in the number of high exposures to
315
SEIBERT
ET AL.
vol. 3, No. 4, April 1996
Adult portable chest Second half, 1994 6oo
f ~,f~
#exams 4661 total exams
23,1%
73,5%
~
5oo 4oo 3oo
Target
-- exposure range 3.4%
Q3-Q4....
200 lOO o
700 550 450 375 325 275 225 175 125 75 <50 Sensitivity number Low 4
Incident Exposure
~ High
FIGURE 3. Adult portable chest examinations for the second half of 1994 are consolidated by quarter (July through September = quarter 3 [Q3] and October through December = quarter 4 [Q4]). The total number of films and percentages within each exposure range are listed at the inside top of the plot.
7.8% occurred simultaneously with an increase in the number of low exposures to 38.3% of the total films. Complaints by the radiologists about noisy films caused by excessive quantum mottle (underexposed technique) provided an educational opportunity for the need to properly expose the computed radiography imaging plate. On the basis of this feedback, a tighter distribution within a range of sensitivity numbers of approximately 125-275 (corresponding to exposures of 1.6-0.7 mR [4.128 × 10-7 C/kg to 1.806 x 10 - 7 C/kg]) was achieved in quarter 3 (Q3) and repeated in quarter 4 (Q4) of 1994, as shown in Figure 3. A reduction in both the low-exposure range to 23.1% and the highexposure range to 3.4% of the total was achieved, with 73.5% of all films within the desired exposure range. Since then, a majority of the portable chest images have been acquired within this exposure range, as shown by the Q4 1994 results presented in Figure 3. Consistent results were obtained through July 1995. Image quality with respect to noise levels has stabilized, and radiologists' satisfaction with the computed radiography images has improved, although still not to the level of optimal screen-film image quality. DISCUSSION
Computed radiography proponents initially promised lower film retakes by virtually eliminating errors caused by under- and overexposure, because of the wide dynamic range characteristics of the detector, and the ability to digitally compensate for variations not possible
316
with screen-film detectors. This part of the promise has been verified by our results and others [10] with our retake rate reduced from 5% to about 1%. Indeed, it also was thought that computed radiography could easily achieve substantial lowering of the radiation dose required for a given examination. Quantitative analysis of computed radiography image detectors by objective performance standards demonstrated a nearly equivalent performance of standard resolution imaging plates with conventional screen-film detectors [11, 12]. Various investigators also have demonstrated the ability to use lower exposures with computed radiography in specific imaging applications such as musculoskeletal [3] and neonatal [5] radiology. However, these findings should be viewed in the context of the conventional screen-film detector speed. A 400-speed system, for example, requires approximately 0.3-mR (0.774 x 10 - 7 C/kg) to 0.5-mR (1.29 x 10 - 7 C/kg) incident exposure to achieve a proper optical density on the processed film, whereas a 200-speed system requires approximately 1 mR. By the same token, others have indicated a requirement for more exposure [7, 8] when compared with a 600- or 400speed system, particularly in instances that require a high-contrast sensitivity. Our results, in general, qualitatively indicate the need for approximately twice the incident exposure on the detector compared with a 400speed system. We are using the computed radiography system for portable radiographic applications without the use of an antiscatter grid, which means that the patient dose and the radiographic techniques are approximately the same as the conventional screen-film detector when using an antiscatter grid having a bucky factor of approximately 2. From our radiologists' perspective, the perceived noise in the 200-speed computed radiography image is comparable to a 400-speed screen-film image. Reduced exposures with computed radiography can be achieved; however, the "acceptable" noise levels desired by our radiologists to comfortably read the image necessitates a higher exposure. For examinations that do not require high-contrast sensitivity (e.g., an examination to identify a naso~astric tube placement), exposure reduction is possible. From the technologists' perspective, the capabilities of the computed radiography system to compensate for under- or overexposure initially caused the perception that a proper radiographic technique was related only to patient positioning because the correct optical density range was obtained on the laser-printed film independent of the technique. In their minds, the paradigm
Vol. 3, No. 4, April 1996
COMPUTED RADIOGRAPHY X-RAY EXPOSURE TRENDS
of screen-film under- or overexposure manifested by corresponding too-light or too-dark films was no longer operative. In our busy institution, experience and educational feedback on the practical advantages and limitations of the computed radiography system took approximately 3-6 months before uniform and optimal exposures were consistently obtained by a majority of the technologists, and then only with identification of the problem and frequent oversight. Increased exposure requirements to the computed radiography imaging plate (relative to a 400-speed screen-film system) are explained by lower X-ray detection efficiency and additional noise sources occurring during the readout phase. Detection efficiency of the computed radiography plate compared with a rareearth phosphor is less because of less overall mass thickness (e.g., -100 mg/cm 2 versus N120 mg/cm 2) and a lower atomic number (z) of the main absorber (barium, z = 56; gadolinium, z = 64). This difference is exacerbated with higher peak kilovoltage examinations such as chest examinations. The noise in the computed radiography image is a contribution of quantum mottle attributable to X-rays, fluctuation of the photostimulated luminescence of the imaging plate, and electronic noise of the analog-to-digital conversion process. Any decrease in the incident exposure requires amplification of the signal to achieve the correct optical density. The net result is the increase of both the quantum and luminance noise. Luminance noise can contribute a significant fraction of the overall noise, particularly in lowexposure situations [2]. There is a possibility that the reported sensitivity values are inaccurate because of miscalibration of the computed radiography reader or changes in the X-ray energy spectrum incident on the detector. Periodic quality-control tests are necessary to validate the sensitivity number accuracy. Patient size variations.and kilovoltage techniques can cause a variation in the effective energy of the transmitted beam, potentially causing a shift of sensitivity numbers to higher or lower values for a given incident exposure [7]. In addition, the method of calculating the sensitivity number depends on the imaging plate readout method. Three different scaling algorithms are used by the Fuji computed radiography system: automatic, semiautomatic, and fixed modes. The automatic m o d e is used most frequently in our system, in which the sensitivity number is determined from an analysis of the signal over the total image area [1, 13]. The semiautomatic mode uses a predefined small area (or areas) over the image (e.g.,
within a 10 × 10 cm centered area of the detector) to determine the sensitivity number. The fixed m o d e does not scale the image information; therefore, the sensitivity number is relevant only if the optical density on the film is within the correct range. Finally, if the signalfinding algorithm fails because of operator or system error [14], the sensitivity number will likely not correlate with incident exposure. Use of the sensitivity number should therefore be considered in terms of the potential inaccuracies; however, a properly calibrated and functioning system will provide the capability to estimate incident exposure within 20%. In summary, computed radiography has provided excellent repeatability of film optical density and has significantly lowered the film repeat rate of the adult portable chest examinations at UCDMC from approximately 5% with conventional screen-film to about 1% with computed radiography. The majority of the retakes are attributable to positioning errors or patient motion. Our findings indicate that computed radiography does not provide the ability to lower patient dose in most imaging examinations (relative to a 400-speed system) because of excessive quantum mottle and a lack of contrast sensitivity unacceptable to the radiologist. This is contrary to the initial expectations of computed radiography and is due to lower X-ray quantum detection efficiency as well as luminance and electronic noise sources added to the signal during readout of the imaging plate. As a general rule, the optimal exposures with computed radiography standard imaging plates (image quality per unit radiation dose) are achieved with approximately 1.5-2 times the incident detector exposure required by a conventional 400-speed rare-earth screen-film combination, equivalent to a sensitivity number range of approximately 125-275. If a grid is not used in conjunction with the computed radiography imaging plate, then the radiographic techniqhe is approximately the same as a 400-speed system with a grid (as is the patient dose). This relates to a grid bucky factor of approximately 2 for the conventional screenfilm detector. Misuse of the auto~a'ffc exisosure ranging capability of computed radiography can cause either a higher patient dose or an increase in image noise and loss of low-contrast sensitivity. Either situation is unacceptable in terms of benefit versus risk of radiation exposure in patient care. Because a proper radiographic technique can be tracked for quality-control evaluation of computed radiography images, the sensitivity number should always be printed with the hardcopy film or soft-copy display. The sensitivity number 317
S E I B E R T ET AL.
Vol. 3, No. 4, April 1996
can be used as an indicator of the relative quantum noise in the image for a given study, but it should not be used as an absolute guide regarding image quality or radiation exposure to the patient.
6.
7. 8.
REFERENCES 1. Kate H. Photostimulable phosphor radiography design considerations. In Seibert JA, Barnes GT, Gould RG, eds. Specification, acceptance testing and quality control of diagnostic X-ray imaging equipment. Woodbury, NY: American Institute of Physics, 1994:731-770. 2. Barnes GT. Digital X-ray image capture with image intensifier and storage phosphor plates: imaging principles, performance and limitations. In Hendee WR, Trueblood JH, eds. Digital imaging. Madison, Wh Medical Physics Publishing, 1993:25-48. 3. Murphey MD, Quale JL, Martin NL, Bramble JM, Cook LT, Dwyer SJ III. Computed radiography in musculoskeletal imaging: state of the art. AJR 1992;158:19-27. 4. Wilson AJ, Mann FA, Murphy WA Jr, Monsees BS, Linn MR. Photostimulable phosphor digital radiography of the extremities: diagnostic accuracy compared with conventional radiography.A JR 1991;157:533-538. 5. Broderick NJ, Long B, Dreesen RG, et al. Phosphor plate computed radi-
9.
10. 11. 12.
13. 14.
ography: response to variation in mAs at fixed kVp in an animal model. Potential role in neonatal imaging. Clin Radio11993;47:39-45. Stringer DA, Cairns RA, Poskitt KJ, Bray H, Milner R, Kennedy B. Comparison of stimulable phosphor technology and conventional screen-film technology in pediatric scoliosis. Pediat Radio11994;24:1-5. Huda W, Arreola M, Jing Z. Computed radiography acceptance testing. SPIE 1995;2342:512-552. Willis CE, Leckie RG, Carter JR, Williamson MP, Scotti SD, Norton G. Objective measures of quality assurance in a computed radiographybased radiology department. SPIE 1995;2342:588-599. Seibert JA. Photestimulable phosphor system acceptance testing. In Seibert JA, Barnes G,l,iGould RG, eds. Specification, acceptance testing and quality control of diagnostic X-ray imaging equipment. Woodbury, NY: American Institute of Physics, 1994:771-800. Sagel SS, Jest RG, Glazer HS, et al. Digital mobile radiography. J Thorac Imaging 1990;5:36-48. Workman A, Cowen AR. Signal, noise and SNR transfer properties of computed radiography. Phys Med Bio11993;38:1789-1808. Sanada S, Doi K, Xu XW, Yin FF, Giger ML, MacMahon H. Comparison of imaging properties of a computed radiography system and screen-film systems. Med Phys 1991 ;18:414-420. Freedman M, Steller D, Jafroudi H, Mun SK. Quality control of storage phosphor digital radiography systems. J Digit Imaging 1995;8:67-74. Solomon SL, Jest RG, Glazer HS, Sagel SS, Anderson DJ, Molina PL Artifacts in computed radiography. AJR 1991;157:181-185.
Announcement T h e N e w Y o r k R o e n t g e n Society is sponsoring a spring conference on May 1-4, 1996, at the Grand Hyatt Hotel in New York City. This 4-day conference is designed for radiologists, radiation oncologists, and other physicians interested in imaging. A spectrum of subspeciality areas will be discussed, and topics of current interest will be presented. The faculty of experts includes many members of the New York Roentgen Society as well as visiting faculty. The fee for practicing physicians is $375; for n o n m e m b e r residents, the fee is $50 with a letter from the chief of service. The fee includes breakfast each morning. Category 1 credit will be awarded as follows: 23 hours for diagnostic radiology and 14.5 hours for radiation oncology. The course directors are Amy Beth Goldman, MD, and Henry A. Pritzker, MD. For more information, contact Henry A. Pritzker, MD, Program Director, Department of Radiology, Montefiore Medical Center, 111 E. 210th St., Bronx, NY 10467; (718) 920-4865, tax (718) 920-4854. Donita Higby also can be contacted at (203) 318-0235.
318
Rationale and Objectives. Computed radiography provides correct optical density on film, independent of the incident radiation exposure, but it can result in under- or overexposure of the imaging plate. In the current study, we evaluated the radiation exposure trends of computed radiography over a 2-year period for portable chest examinations to determine and compare the radiographic techniques of the computed radiography system relative to conventional screen-film detectors. Methods. A Fuji computed radiography system was interfaced to a digital workstation to track system usage and examination demographics, including examination type and sensitivity number. Hard-copy films were used for diagnosis. The sensitivity number, a value inversely related to incident exposure on the imaging plate, was used to determine whether the proper techniques were used by the technologists. Results. The initial use of the computed radiography system revealed a broad distribution of exposures being used; complaints regarding noisy films (e.g,, underexposure) resulted in subsequent overexposure for a significant number of films. A quality-control audit indicating excessive exposure resulted in educational feedback and a tighter distribution of exposures within the optimal range as determined by our radiologists, The average technique was approximately equivalent to a 200-speed system.
Received September 18, 1995, and accepted for publication after revision November 29, 1995.
Conclusion. Computed radiography provides excellent dynamic range and rescaling capabilities for proper film optical density, and th~_ts fewer repeat examinations. However, underexposure results in suboptimal image quality that is related to excessive quantum mottle. Overexposure requires film audits to limit unnecessary radiation exposure. In general, the optimal e:kposures are achieved with approximately 1.5=2, times the incident detector exposure of a 400-speed rare-earth system. The ability of computed radiography to reduce radiation exposure is unlikely when compared with a typical rare-earth screen-film combination (400 speed) in terms of adequate image quality for the diagnosis of subtle, low-contrast findings. For certain diagnostic procedures (e.g., nasogastric tube placement verification), lower exposures can be tolerated.
Acad Radio/1996;3.'313-318 © 1996, Association of University Radiologists
Key Words. Computed radiography; radiation dose; portable chest imaging; photostimulable phosphor detectors; digital X-ray imaging.
From the Department of Radiology, University of California Davis Medical Center, Sacramento, CA. Parts of this article were presented at the 43rd Annual Meeting of the Association of University Radiologists held in San Diego, CA, in May 1995. Address reprint requests to J. A. Seibert, PhD, Radiology Research, University of California Davis Medical Center, FOLB-II-E, 2421 45th St., Sacramento, CA 95817.
313
SEIBERT
ET AL.
omputed radiography is the name given to the pro-
cess of digital image acquisition using an X-ray C detector comprised of a photostimulable phosphor imaging plate and an image reader-writer that processes the latent image information for subsequent brightness scab ing and laser printing on film [1]. Separation of the acquisition and the image display events using digital acquisition and processing of the signal permits the correct optical density range to be achieved in the final image, independent of the exposure to the imaging plate. This is attributed to a linear signal response over four orders of magnitude (i.e., 104 or 10,000 times) in incident exposure. An internal scaling algorithm within the computed radiography reader corrects for under- or overexposure caused by inappropriate radiographic techniques. This scaling algorithm also remaps the variations in exposure in a manner similar to a film characteristic curve response and provides a consistent optical density on the printed film. These are significant benefits of computed radiography compared with screen-film systems. On the other hand, limiting the spatial resolution of the computed radiography image is much less than screen-film, and the X-ray detection efficiency of the imaging plate is less than a corresponding 400-speed rare-earth screen. In addition, secondary noise sources caused by luminance variations during the readout process [2], as well as electronic signal variations during digitization, increase the perceptible image noise for a given incident exposure. With current technology, the images are most often presented with a reduced film format size, making comparisons to full-size films suboptimal. Despite these drawbacks, improvements in technology, decreasing system costs, and a tangible future for electronic imaging in medicine have spurred an interest in clinical implementation of such systems. With expanding use, computed radiography is expected to be the frontline technique for acquiring digital large-area projection images in the coming years. It is therefore important to understand the features and limitations of these systems for X-ray projection imaging. There are reports demonstrating the ability to reduce exposures with computed radiography compared with conventional screen-film imaging without losing diagnostic accuracy [3, 4]. Exposure reductions of up to 50% have been claimed, although the speed of the screenfilm detector was not specified [3]. Comparison with a 200-speed system for pediatric applications also indicated an ability to reduce exposures [5, 6]. Other researchers have reported an increase in exposures
314
VoJ. 3, No. 4, April 1996
required with-computed radiography relative to a 400or 600-speed conventional screen-film system [7, 8]. A common theme to all of the investigations is the "potential" to reduce exposure by computed radiography depending on the examination type and the diagnosis requested. On the basis of this background, the purpose of this study was to evaluate the radiation exposure trends of computed radiography over a 2year period for adt~llt portable chest examinations at the University of California Davis Medical Center (UCDMC) and to compare the results with a conventional 400speed screen-film detector. We selected adult portable chest examinations as the focus because of the large number of examinations and the large number of repeat examinations, with the majority being caused by under- or overexposure (i.e., films were too light or too dark for optimal diagnosis). Initially, we thought that computed radiography would be an uncompromising solution to many of the problems encountered with suboptimal radiographic technique encountered with screen-film detectors. Because under- or overexposure is not readily apparent on the resultant computed radiography film images, standardization of the radiographic technique to determine the exposure range that provides reasonably lowqmage noise at a minimum exposure level without compromising the quality of the image was the goal of our study. MATERIALS AND METHODS
A calibrated Fuji computed radiography system (model AC-1 plus; Fuji, Tokyo, Japan) using standard resolution (ST-III-N) imaging plates produced clinical portable chest images on laser-printed film. Over the 2-year period of the study, approximately 30% of all adult portable chest images were acquired with computed radiography. Typical parameters used by the technologists were 85 kVp and 5-10 mAs and no antiscatter grid. The processing algorithm for the portable chest image was modified to print a single 9 x 11 inch (23 x 28 cm) image on reduced-format film (a two-thirds reduction factor of a standard 14 x 17 inch [35 x 43 cm] film). Over an 18month period, from July 1993 through December 1994, computed radiography examinations (both image and demographic information) were automatically transferred to a computer workstation via a digital acquisition system interface. The digital (soft-copy) image was not used for viewing or diagnosis. Information was collected regarding exam type, sensitivity number, plate size, and image latitude for all computed radiography examina-
Vol. 3, No. 4, April 1996
COMPUTED
tions, and stored on a month by month basis. Adult portable chest examination information was sorted from all other studies using a personal computer spreadsheet program. The sensitivity number was used to estimate the average incident exposure on the imaging plate for each image as follows: incident exposure (mR) =
200 sensitivity number
where 1 mR = 2.58 x 10-7 C/kg air. Thus, a sensitivity number of 200 indicates an approximate incident exposure of 1 mR (2.58 x 10-7 C/kg) to the imaging plate, whereas a sensitivity number of 600 represents an incident exposure of approximately 0.3 mR (0.774 x 10 - 7 C/kg). Verification of the sensitivity number accuracy was accomplished during the acceptance testing and subsequent quality-control checks on the computed radiography unit using a defined quality-control methodology [9]. Any retake images were consolidated with the total number of computed radiography films evaluated, and these represented approximately 1% of the total, based on a retake rate analysis. (Conventional screen-film images were not included in this evaluation.) Sensitivity numbers above 20,000 were excluded from the data set as u n e x p o s e d imaging plates. Incident exposure estimates were plotted as a function of the number of examinations (i.e., a histogram). Quarterly reports were presented to a radiology quality improvement committee for analysis, review, and determination of exposure trends for portable chest imaging using computed radiography. An opportunity for educational feedback to the technologists was channeled through the pulmonary radiologists, who qualitatively rendered a judgment about the film quality regarding perceived image noise. This communication was used in conjunction with image audits of the computed radiography data files to determine the "optimal" image quality in terms of sensitivity numbers (incident exposure on the detector), as tracked over the evaluation period. These exposure limits, were then used retrospectively on the 18 months of acquired exposure data. RESULTS
The initial use of the computed radiography system (over the first year of evaluating sensitivity numbers) revealed that there was a broad distribution of exposures being used in portable chest examinations, as shown in Figure 1 for the second half of 1993. The incident exposures estimated from the sensitivity numbers
RADIOGRAPHY
X-RAY
EXPOSURE
Adult portable chest Second half, 1993
~--
#exams 600
4824 total exams
58.1%
30.1%
TRENDS
Target exposure range 11.8%
500
Q3-400 Q4 . m 300 200
,,o.
100 0
700 550 450 375 325 2
225 175 125 75 <50
Sensitivity number Low -
Incident Exposure
= High
FIGURE 1. The number of portable chest X-ray examinations plotted as a function of incident exposure to the computed radiography imaging plate as determined by the sensitivity number. Adult portable chest examinations for the second half of 1993 are consolidated by quarter (July to September = quarter 3 [Q3] and October to December = quarter 4 [Q4]). Each point on the c u r v e represents a range of sensitivity values shown by the x-axis labels (not all points are labeled). The total number of films and percentages within each exposure range are listed at the inside top of the plot.
Adult portable chest First
half,1994
Target exposure range
~--
#exams 600
4572 total exams
38.3%
53.9%
7.8%
500
400
QI m
300
Q2 d
20O
.ir/Q~,
IO0 0
./J~ .,e"*')
700 580 450 375 325 2:
~-,..z% 225 175 125 75 <50
Sensitivity number Low "=
Incident Exposure
=- High
FIGURE 2. Adult portable chest examinations for the first half of 1994 ~.re consolidated by quarter (January through March = quarter 1 [Q1] and April through June = quarter 2 [Q2]). The total number of films and percentages within each exposure range are listed at the inside top of the plot.
were uniform across a broad s p e c t r u m / w i t h 30.1% of the films falling in the low (sensitivity > 275) and 11.8% of films falling in the high (sensitivity < 125) exposure ranges. During the first quarter of 1994, the technologists were told of the exposure evaluation study and the presence of the large number of high-incident exposures (low sensitivity numbers). A trend toward lower exposures occurred in quarter 2 (Q2), as illustrated by the comparison of Q1 and Q2 curves in Figure 2. A decrease in the number of high exposures to
315
SEIBERT
ET AL.
vol. 3, No. 4, April 1996
Adult portable chest Second half, 1994 6oo
f ~,f~
#exams 4661 total exams
23,1%
73,5%
~
5oo 4oo 3oo
Target
-- exposure range 3.4%
Q3-Q4....
200 lOO o
700 550 450 375 325 275 225 175 125 75 <50 Sensitivity number Low 4
Incident Exposure
~ High
FIGURE 3. Adult portable chest examinations for the second half of 1994 are consolidated by quarter (July through September = quarter 3 [Q3] and October through December = quarter 4 [Q4]). The total number of films and percentages within each exposure range are listed at the inside top of the plot.
7.8% occurred simultaneously with an increase in the number of low exposures to 38.3% of the total films. Complaints by the radiologists about noisy films caused by excessive quantum mottle (underexposed technique) provided an educational opportunity for the need to properly expose the computed radiography imaging plate. On the basis of this feedback, a tighter distribution within a range of sensitivity numbers of approximately 125-275 (corresponding to exposures of 1.6-0.7 mR [4.128 × 10-7 C/kg to 1.806 x 10 - 7 C/kg]) was achieved in quarter 3 (Q3) and repeated in quarter 4 (Q4) of 1994, as shown in Figure 3. A reduction in both the low-exposure range to 23.1% and the highexposure range to 3.4% of the total was achieved, with 73.5% of all films within the desired exposure range. Since then, a majority of the portable chest images have been acquired within this exposure range, as shown by the Q4 1994 results presented in Figure 3. Consistent results were obtained through July 1995. Image quality with respect to noise levels has stabilized, and radiologists' satisfaction with the computed radiography images has improved, although still not to the level of optimal screen-film image quality. DISCUSSION
Computed radiography proponents initially promised lower film retakes by virtually eliminating errors caused by under- and overexposure, because of the wide dynamic range characteristics of the detector, and the ability to digitally compensate for variations not possible
316
with screen-film detectors. This part of the promise has been verified by our results and others [10] with our retake rate reduced from 5% to about 1%. Indeed, it also was thought that computed radiography could easily achieve substantial lowering of the radiation dose required for a given examination. Quantitative analysis of computed radiography image detectors by objective performance standards demonstrated a nearly equivalent performance of standard resolution imaging plates with conventional screen-film detectors [11, 12]. Various investigators also have demonstrated the ability to use lower exposures with computed radiography in specific imaging applications such as musculoskeletal [3] and neonatal [5] radiology. However, these findings should be viewed in the context of the conventional screen-film detector speed. A 400-speed system, for example, requires approximately 0.3-mR (0.774 x 10 - 7 C/kg) to 0.5-mR (1.29 x 10 - 7 C/kg) incident exposure to achieve a proper optical density on the processed film, whereas a 200-speed system requires approximately 1 mR. By the same token, others have indicated a requirement for more exposure [7, 8] when compared with a 600- or 400speed system, particularly in instances that require a high-contrast sensitivity. Our results, in general, qualitatively indicate the need for approximately twice the incident exposure on the detector compared with a 400speed system. We are using the computed radiography system for portable radiographic applications without the use of an antiscatter grid, which means that the patient dose and the radiographic techniques are approximately the same as the conventional screen-film detector when using an antiscatter grid having a bucky factor of approximately 2. From our radiologists' perspective, the perceived noise in the 200-speed computed radiography image is comparable to a 400-speed screen-film image. Reduced exposures with computed radiography can be achieved; however, the "acceptable" noise levels desired by our radiologists to comfortably read the image necessitates a higher exposure. For examinations that do not require high-contrast sensitivity (e.g., an examination to identify a naso~astric tube placement), exposure reduction is possible. From the technologists' perspective, the capabilities of the computed radiography system to compensate for under- or overexposure initially caused the perception that a proper radiographic technique was related only to patient positioning because the correct optical density range was obtained on the laser-printed film independent of the technique. In their minds, the paradigm
Vol. 3, No. 4, April 1996
COMPUTED RADIOGRAPHY X-RAY EXPOSURE TRENDS
of screen-film under- or overexposure manifested by corresponding too-light or too-dark films was no longer operative. In our busy institution, experience and educational feedback on the practical advantages and limitations of the computed radiography system took approximately 3-6 months before uniform and optimal exposures were consistently obtained by a majority of the technologists, and then only with identification of the problem and frequent oversight. Increased exposure requirements to the computed radiography imaging plate (relative to a 400-speed screen-film system) are explained by lower X-ray detection efficiency and additional noise sources occurring during the readout phase. Detection efficiency of the computed radiography plate compared with a rareearth phosphor is less because of less overall mass thickness (e.g., -100 mg/cm 2 versus N120 mg/cm 2) and a lower atomic number (z) of the main absorber (barium, z = 56; gadolinium, z = 64). This difference is exacerbated with higher peak kilovoltage examinations such as chest examinations. The noise in the computed radiography image is a contribution of quantum mottle attributable to X-rays, fluctuation of the photostimulated luminescence of the imaging plate, and electronic noise of the analog-to-digital conversion process. Any decrease in the incident exposure requires amplification of the signal to achieve the correct optical density. The net result is the increase of both the quantum and luminance noise. Luminance noise can contribute a significant fraction of the overall noise, particularly in lowexposure situations [2]. There is a possibility that the reported sensitivity values are inaccurate because of miscalibration of the computed radiography reader or changes in the X-ray energy spectrum incident on the detector. Periodic quality-control tests are necessary to validate the sensitivity number accuracy. Patient size variations.and kilovoltage techniques can cause a variation in the effective energy of the transmitted beam, potentially causing a shift of sensitivity numbers to higher or lower values for a given incident exposure [7]. In addition, the method of calculating the sensitivity number depends on the imaging plate readout method. Three different scaling algorithms are used by the Fuji computed radiography system: automatic, semiautomatic, and fixed modes. The automatic m o d e is used most frequently in our system, in which the sensitivity number is determined from an analysis of the signal over the total image area [1, 13]. The semiautomatic mode uses a predefined small area (or areas) over the image (e.g.,
within a 10 × 10 cm centered area of the detector) to determine the sensitivity number. The fixed m o d e does not scale the image information; therefore, the sensitivity number is relevant only if the optical density on the film is within the correct range. Finally, if the signalfinding algorithm fails because of operator or system error [14], the sensitivity number will likely not correlate with incident exposure. Use of the sensitivity number should therefore be considered in terms of the potential inaccuracies; however, a properly calibrated and functioning system will provide the capability to estimate incident exposure within 20%. In summary, computed radiography has provided excellent repeatability of film optical density and has significantly lowered the film repeat rate of the adult portable chest examinations at UCDMC from approximately 5% with conventional screen-film to about 1% with computed radiography. The majority of the retakes are attributable to positioning errors or patient motion. Our findings indicate that computed radiography does not provide the ability to lower patient dose in most imaging examinations (relative to a 400-speed system) because of excessive quantum mottle and a lack of contrast sensitivity unacceptable to the radiologist. This is contrary to the initial expectations of computed radiography and is due to lower X-ray quantum detection efficiency as well as luminance and electronic noise sources added to the signal during readout of the imaging plate. As a general rule, the optimal exposures with computed radiography standard imaging plates (image quality per unit radiation dose) are achieved with approximately 1.5-2 times the incident detector exposure required by a conventional 400-speed rare-earth screen-film combination, equivalent to a sensitivity number range of approximately 125-275. If a grid is not used in conjunction with the computed radiography imaging plate, then the radiographic techniqhe is approximately the same as a 400-speed system with a grid (as is the patient dose). This relates to a grid bucky factor of approximately 2 for the conventional screenfilm detector. Misuse of the auto~a'ffc exisosure ranging capability of computed radiography can cause either a higher patient dose or an increase in image noise and loss of low-contrast sensitivity. Either situation is unacceptable in terms of benefit versus risk of radiation exposure in patient care. Because a proper radiographic technique can be tracked for quality-control evaluation of computed radiography images, the sensitivity number should always be printed with the hardcopy film or soft-copy display. The sensitivity number 317
S E I B E R T ET AL.
Vol. 3, No. 4, April 1996
can be used as an indicator of the relative quantum noise in the image for a given study, but it should not be used as an absolute guide regarding image quality or radiation exposure to the patient.
6.
7. 8.
REFERENCES 1. Kate H. Photostimulable phosphor radiography design considerations. In Seibert JA, Barnes GT, Gould RG, eds. Specification, acceptance testing and quality control of diagnostic X-ray imaging equipment. Woodbury, NY: American Institute of Physics, 1994:731-770. 2. Barnes GT. Digital X-ray image capture with image intensifier and storage phosphor plates: imaging principles, performance and limitations. In Hendee WR, Trueblood JH, eds. Digital imaging. Madison, Wh Medical Physics Publishing, 1993:25-48. 3. Murphey MD, Quale JL, Martin NL, Bramble JM, Cook LT, Dwyer SJ III. Computed radiography in musculoskeletal imaging: state of the art. AJR 1992;158:19-27. 4. Wilson AJ, Mann FA, Murphy WA Jr, Monsees BS, Linn MR. Photostimulable phosphor digital radiography of the extremities: diagnostic accuracy compared with conventional radiography.A JR 1991;157:533-538. 5. Broderick NJ, Long B, Dreesen RG, et al. Phosphor plate computed radi-
9.
10. 11. 12.
13. 14.
ography: response to variation in mAs at fixed kVp in an animal model. Potential role in neonatal imaging. Clin Radio11993;47:39-45. Stringer DA, Cairns RA, Poskitt KJ, Bray H, Milner R, Kennedy B. Comparison of stimulable phosphor technology and conventional screen-film technology in pediatric scoliosis. Pediat Radio11994;24:1-5. Huda W, Arreola M, Jing Z. Computed radiography acceptance testing. SPIE 1995;2342:512-552. Willis CE, Leckie RG, Carter JR, Williamson MP, Scotti SD, Norton G. Objective measures of quality assurance in a computed radiographybased radiology department. SPIE 1995;2342:588-599. Seibert JA. Photestimulable phosphor system acceptance testing. In Seibert JA, Barnes G,l,iGould RG, eds. Specification, acceptance testing and quality control of diagnostic X-ray imaging equipment. Woodbury, NY: American Institute of Physics, 1994:771-800. Sagel SS, Jest RG, Glazer HS, et al. Digital mobile radiography. J Thorac Imaging 1990;5:36-48. Workman A, Cowen AR. Signal, noise and SNR transfer properties of computed radiography. Phys Med Bio11993;38:1789-1808. Sanada S, Doi K, Xu XW, Yin FF, Giger ML, MacMahon H. Comparison of imaging properties of a computed radiography system and screen-film systems. Med Phys 1991 ;18:414-420. Freedman M, Steller D, Jafroudi H, Mun SK. Quality control of storage phosphor digital radiography systems. J Digit Imaging 1995;8:67-74. Solomon SL, Jest RG, Glazer HS, Sagel SS, Anderson DJ, Molina PL Artifacts in computed radiography. AJR 1991;157:181-185.
Announcement T h e N e w Y o r k R o e n t g e n Society is sponsoring a spring conference on May 1-4, 1996, at the Grand Hyatt Hotel in New York City. This 4-day conference is designed for radiologists, radiation oncologists, and other physicians interested in imaging. A spectrum of subspeciality areas will be discussed, and topics of current interest will be presented. The faculty of experts includes many members of the New York Roentgen Society as well as visiting faculty. The fee for practicing physicians is $375; for n o n m e m b e r residents, the fee is $50 with a letter from the chief of service. The fee includes breakfast each morning. Category 1 credit will be awarded as follows: 23 hours for diagnostic radiology and 14.5 hours for radiation oncology. The course directors are Amy Beth Goldman, MD, and Henry A. Pritzker, MD. For more information, contact Henry A. Pritzker, MD, Program Director, Department of Radiology, Montefiore Medical Center, 111 E. 210th St., Bronx, NY 10467; (718) 920-4865, tax (718) 920-4854. Donita Higby also can be contacted at (203) 318-0235.
318