Energy dependence of photostimulable phosphor. (Computed Radiography).
Given this linearity, one could infer the amount of radiation exposure from the stimulated luminescence. However, the response can be diminished by a delay in processing (3) and depends on the beam quality of the radiation. (4-6) Huda et al (4) studied the energy dependence of the PSP's response to the broad x-ray spectrum in a clinical environment. They used acrylic phantoms for their investigation. With the primary objective of comparing PSP with film-screens, they expressed their findings relative to film-screen systems. The latter also are energy dependent, especially at low energy.
In calibrating an imaging system, it would be more helpful to describe the energy dependence in terms of readily measurable parameters, such as index of exposure or sensitivity displayed by the image reader and the radiation recorded in an ion chamber. Tucker and Rezentes (5) presented the relationship between the response and beam quality in these terms. However, beam qualities achieved by inserting copper filters are not the same as the exit radiation from human bodies where there is scatter radiation.
Most recently, Christodoulu et al (6) showed the energy dependence of the image plate reader's sensitivity with phantoms of rectangular acrylic sheets or blocks. Measurements at 70 kVp were made with 10.16 cm (4 inches) and 15.24 cm (6 inches) of acrylic. Measurements at 90 kVp and 125 kVp were made with 10.16 cm (4 inches), 15.24 cm (6 inches) and 20.32 cm (8 inches) of acrylic. The results were normalized to the same amount of exit exposure.
To illustrate the dependence of the phosphor's response on the quality of radiation exiting a patient, we performed similar measurements but with combinations of acrylic sheets and aluminum sheets to represent an extremity, a skull, an abdomen and a chest. For each phantom, images were taken at several different values of kVp. The results were compared at the same amount of air kerma (exposure) at the imaging plate when possible. For the chest phantom, this was 15.38 [micro]Gy (1.77 mR). For all other procedures, it was 9.27 [micro]Gy (1.07 mR).
Methods and Materials
The computed radiography system in the authors' clinic (Fuji Model FCR 9000, Fuji Photo Film, Tokyo, Japan) uses photostimulable phosphor imaging plates (model ST-VA). The compound, BaF ([Br.sub.85%][I.sub.15%]), is used as the phosphor and has a primary radiation absorption K-edge of' barium at 37.4 keV and a much smaller absorption K-edge of iodine at 13.4 keV. (4,5) The radiographic machine used in this study (model MPX, General Electric Medical Systems, Milwaukee, Wis) was calibrated with an accuracy of 5% or better from 50 to 120 kVp. The 3-phase high voltage had a ripple of 5%. The anode angle was 12 [degrees]. The beam quality of the incident radiation was characterized by half:value layers of 1.7 mm, 2.8 mm, 3.2 mm and 3.8 mm of aluminum for 50 kVp, 80 kVp, 95 kVp and 110 kVp, respectively.
Two ion chambers of the same type were used in this investigation (model 10X5-6, Radcal, Monrovia, Calif). The-ion chambers were simultaneously irradiated 10 times to determine their precision. For each exposure, the ratio of the 2 recorded values was computed. Then the mean and standard deviation of the ratios were calculated. Measurements were made at 50 kVp with nominal exposures of 17.4 [micro]Gy (2 mR), 174 [micro]Gy (20 mR) and 869 [micro]Gy (100 mR). Measurements were repeated at 80 kVp and with approximately 17.4 [micro]Gy (2 mR). Results are presented in Table 1.
Assuming that both detectors had the same reproducibility, the precision deduced from the standard deviations would be smaller than 0.5%. Though both chambers were calibrated by the manufacturer annually, one chamber consistently displayed slightly lower value. This difference could be attributed to differences in calibration or placement of the 2 chambers. In subsequent measurements in which only one chamber was used, slight inaccuracy would lead to a constant overall correction factor, whereas imprecision could affect the determination of energy dependence. The literature from the manufacturer shows that this ion chamber model typically has energy dependence better than 1%. The total uncertainty in an ion chamber measurement due to energy dependence and imprecision was therefore estimated to be approximately 1%.
In this investigation, after an imaging plate was irradiated, it was processed by a test protocol in the "exposure data recognizer" (EDR) of the Fuji image plate reader. Semiautomatic mode with a fixed latitude (L) of 1 was applied to the images. In this semiautomatic mode, the reader analyzed the stimulated luminescence from the central portion (10 cm x 10 cm) of the imaging plate.7 The logarithm of the stimulated luminescence intensity of each pixel in this region of interest (ROI) was tallied.
The L number is an index expressing the selected range of stimulated luminescence intensity (in logarithmic scale). This range of value is mapped to digital output values of 0 to 1023. Thus, L could be considered analogous to the latitude of film-screen systems. A smaller value of L would describe a narrower-latitude imaging system. The median ([S.sub.k]) of the stimulated luminescence intensity in logarithmic scale corresponds to the midvalue of the display scale. [S.sub.k] is used to compute the S-number, an index of reading sensitivity. (7) The final result is a linear but inverse relationship between the S-value and the average air kerma or exposure of the region of interest. (8)
At installation, the computed radiography system was calibrated by the manufacturer to give an S-value of 200 for 8.69 [micro]Gy (1 mR) at 80 kVp with a source-to-image distance (SID) of 1.8 meters and without additional filtration. This calibration for imaging plates of both sizes (24 x 30 cm and 35.4 x 43.0 cm) was verified at the acceptance of the machine.
In a typical clinical environment, the time interval between exposing image plates and feeding them to the image reader varies and cannot be controlled. In this investigation, we studied the fading of the latent image during processing delay by exposing one image plate at 80 kVp at a source-to-image distance of 81 cm and at approximately 12.2 [micro]Gy (1.4 mR), noting the delay time before the imaging plate was processed and recording the corresponding S-value. The same imaging plate was thus irradiated repeatedly and the processing delay was intentionally varied from the minimum of 102 seconds (1 minute, 42 seconds) to 1639 seconds (27 minutes, 19 seconds). Results are presented in Fig. 1.
[FIGURE 1 OMITTED]
To determine typical radiation exposures, this study used modified ANSI diagnostic x-ray phantoms. (9) described in the American Association of Physicists in Medicine Report No. 31. (10) The extremity phantom consisted of sheets of materials with the following thicknesses: 2.54 cm (1 inch) acrylic, 2 mm aluminum and 2.54 cm (1 inch) acrylic. The skull phantom consisted of 2.54 cm (1 inch) clear acrylic, 1 mm of aluminum, 2.54 cm (1 inch) acrylic, 5.08 cm (2 inches) acrylic, 2.54 cm (1 inch) acrylic, 2 mm aluminum and 2.54 cm (1 inch) acrylic. The chest phantom had a similar configuration as the skull phantom, except the 5.08 cm (2 inches) of acrylic was replaced by an air gap of the same thickness. The abdomen phantom was made up of acrylic sheets with 17.78 cm (7 inches) total thickness.
Radiographic exposures in this investigation were made using the following technique. The source-to-image distance (ie, source to imaging plate) was 100 cm. The collimators were opened to cover the full area of the imaging plate. There was an air gap of 8.2 cm between the phantom and the cassette containing an imaging plate. The imaging plate measured 25.4 x 30.5 cm (10 x 12 inches). To measure the exit exposure from the phantom, an ion chamber was placed in the air gap between the phantom and the cassette. This ion chamber was 3.5 cm from the imaging plate and was located at a corner of the imaging plate. Exposure to an imaging plate was computed from these measurements with correction for distances.
The same imaging plate was used for all measurements. For each kVp, the tube current was maintained while the exposure time was varied to give different amounts of radiation. S-values and corresponding exposures were acquired for each kVp. Exposures were made within 10% of the desired values. Using the inverse relation of S-value to exposure, the S-value corresponding to a desired amount of exposure to the imaging plate was computed by interpolation. The desirable value was 8.69 [micro]Gy (1.00 mR), the value used for calibration by the manufacturer. However, for the higher kilovoltage used for chest examinations, with the chosen SID, the minimum achievable exposure was higher. After 7% correction for heel effects at the ion chamber, the chosen reference value of air kerma (exposure) at the center of the imaging plate was 15.38 [micro]Gy (1.77 mR) for the chest phantom and 9.27 [micro]Gy (1.07 mR) for the others. After normalizing the experimental data to the reference value of exposure at the imaging plate, we established the relationship of sensitivity vs kVp for the radiographic examination of the test objects. (See Figs. 2 to 5.)
[FIGURES 2-5 OMITTED]
Figure 1 shows that the S-value increased with delay in processing. That is, the stimulated luminescence, which is inversely related to the S-value, diminished with time, (3) in agreement with Floyd et al. (3) In this investigation, the delay in processing was as long as 7 minutes. Figure 1A shows that the S-value could increase by 4% at the end of this time interval. The effect of a shorter delay (ie, from 102 seconds to 240 seconds) can be seen in Figure 1B. The mean of these measurements has a standard deviation of 1%. This could be interpreted as the reproducibility of an individual imaging plate and is much smaller than the variation introduced by the delay in processing the imaging plates. The combination of the uncertainties in ion chamber measurement, reproducibility of the PSP's response and the delay in processing would be an estimated 5%. An overall uncertainty of this magnitude was applied to the results of this investigation.
S-values obtained at different values of kVp for constant values of air kerma (exposure) are presented in Figures 2 to 5. A chi-square test showed that the measurements are unlikely to be compatible with a model of constant response. (See Table 2.) A regression analysis (SAS Version 6.12, SAS Institute, Cary, NC) yielded the coefficients in a linear model and in a quadratic model. Chi-square tests showed the linear model to be sufficient and the quadratic model might be better in most cases. In other words, the luminescence of the phosphor increases with increasing kVp per unit of exposure or air kerma at the imaging cassette. For the abdomen phantom and the chest phantom, the S-value appears to be less energy dependent above 90 kVp.
Measurements in this experiment were assumed to have the same fractional uncertainty. Tucker and Rezente (5) studied the energy dependence of this parameter. They examined the statistical fluctuation in the sampling luminance. That is, for a uniformly exposed image plate, not all the pixels would yield exactly the same amount of luminance. The standard deviation of this statistical distribution might depend on the incident energy. However, this spectral dependence was negligible compared with the uncertainty introduced in processing delay. The assignment of an overall 5% uncertainty in this investigation is comparable to findings by Floyd et al (3) of 1.6% to 4.2% accuracy in the use of a single imaging plate and 5.0% to 5.9% uncertainty when several plates were involved.
The ranges of energy chosen for this study encompass the radiographic techniques currently used in the authors' facility and the values used by other investigators for the extremities (11) and lungs. (11,12) The range of kilovoltage used for each phantom in this study extended beyond typical values of conventional film-screen systems. This broad exploration could be of value when the image processing ability with computed radiography extends the useable ranges of kilovoltage. The study by Christodoulou et al (6) is most similar to this one. They used acrylic phantoms for measurements at 70 kVp, 90 kVp and 125 kVp. This study used aluminum to simulate bones and acrylic materials for soft tissue, and it covered a wider range of 50 kVp to 120 kVp.
In this study, the air gap between the phantom and imaging plate was used to monitor air kerma (exposure). This geometry could modify slightly the spectrum of radiation impinging on the imaging plate. Measurements of air kerma (exposure) in this experiment included scatter from the phantom as well as from the imaging cassette. Corrections for these secondary effects were not addressed in this exercise. Use of a grid also would change the quality of exit radiation and was not thoroughly explored in this investigation. Within the confines of these uncertainties and limitations, the spectral dependence of luminance on radiation exiting radiographic phantoms has been demonstrated. The general decrease in sensitivity per unit of exposure with kVp agrees qualitatively with Christodoulou et al. (6) A quantitative comparison cannot be made because these 2 experiments employed radiation of different qualities. The study reported here was intended to aid the use of S-value in calibrating computed radiographic systems and interpreting S-value in radiographic examinations. Photon flux or energy fluence per Roentgen or per unit air kerma is highly energy dependent. Photon fluence or flux increases with effective energy until near the upper end of the energy range used in this investigation. Therefore, as the kilovoltage increases, there are more photons or energy per unit of air kerma (or per unit of exposure) in the incident radiation, and it consequently can yield more luminance. Qualitatively, this could explain increasing amounts of luminance at higher kilovoltage reported in this and other studies. (5,6)
Table 1 Comparison of 2 Ion Chambers Irradiated Simultaneously * kVp Nominal Air Kerma Ratio of Chamber 1 to (Exposure) Chamber 2 (Mean [+ or -] Standard Deviation) 50 17.4 [micro]Gy (2 mR) 0.928 [+ or -] 0.006 50 174 [micro]Gy (20 mR) 0.940 [+ or -] 0.002 50 869 [micro]Gy (100 mR) 0.934 [+ or -] 0.004 80 0.926 [+ or -] 0.006 * Ten measurements were made at a technique and nominal air kerma (exposure). Table 2 Probability of Chi-Squared Fit of Response (S-value vs Kilovoltage at Constant Exposure) To Different Models Radiographic Model Technique Constant Linear Quadratic Extremity 0.00 0.89 0.99 Skull 0.02 0.89 0.84 Abdomen 0.02 0.91 1.00 Chest 0.24 0.95 1.00
(1.) Sonoda M, Takano M, Migahara J, Kato H. Computed radiography utilizing scanning laser stimulated luminescence. Radiology. 1983;148:833-838.
(2.) Cowen AR, Workman A, Price JS. Physical aspects of photostimulable phosphor computed radiography. Br J Radiol. 1993;66:332-345.
(3.) Floyd CE, Chotas HG, Dobbins III JT, C. E. Ravin CE. Quantitative radiographic imaging using a photostimulable phosphor system. Med Phys. 1990;17:454-459.
(4.) Huda W, Rill L, Bruner A. Relative speeds of Kodak computed radiography phosphors and screen film systems. Med Phys. 1997;24:1621-1628.
(5.) Tucker D, Rezente P. The relationship between pixel value and beam quality in photostimulable phosphor imaging. Med Phys. 1997;24:887-893.
(6.) Christodoulou EG, Goodsitt MM, Chan H. Phototimer setup for CR imaging. Med Phys. 2000;27: 2652-2658.
(7.) Fuji Computed Radiography Technical Review No. 3. Automatic setting functions for image density and range in the FCR system. Fuji Photo Film Co Ltd. Ref. No. XM-501E (SK. 95.05.MW), 1993.
(8.) 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 Institute of Physics Inc; 1994:771-800. American Association of Physicists in Medicine Medical Physics Monograph No. 20.
(9.) ANSI PH2-43, Method for the sensitometry of medical X-ray screen-film processing systems. New York, NY: American National Standards Institute, 1982.
(10.) Standardized methods for measuring diagnostic X-ray exposures. American Institute of Physics, College Park, Md. 1990. AAPM Report No. 31.
(11.) Nakano Y, Hiraoka T, Togashi K, et al. Direct radiographic magnification with computed radiography. AJR Am J Roentgenol. 1987;148: 569-573.
(12.) Oda N, Nakata H, Murakami S, Terada K, Nakamura K, Yoshida A. Optimal beam quality for chest computed radiography. Invest Radiol. 1996;31:126-131.
Robert Y.L. Chu, Ph.D., is a medical physicist and director of medical physics in the department of radiological sciences at the University of Oklahoma. He is also a medical physicist at the Veterans Affairs Medical Center in Oklahoma City, Okla.
Elizabeth N. Christian, M.S., is a medical physicist at Satilla Regional Cancer Center in Waycross, Ga.
Bob G. Eaton, M.D., is a radiologist and professor emeritus of radiological sciences at the University of Oklahoma.
|Printer friendly Cite/link Email Feedback|
|Author:||Chu, Robert Y.L.; Christian, Elizabeth N.; Eaton, Bob G.|
|Article Type:||Statistical Data Included|
|Date:||Mar 1, 2002|
|Previous Article:||Cystogram. (Patient Page).|
|Next Article:||Diagnosis and treatment of stroke. (CE Directed Reading).|