Effect of patient support pads on image quality and dose in fluoroscopy a)

Effect of patient support pads on image quality and dose in fluoroscopy a) William R. Geiser, Walter Huda, b) and Nikolaos A. Gkanatsios Department of Radiology, University of Florida, Gainesville, Florida 32610-0374 Received 20 September 1996; accepted for publication 10 December 1996 An investigation was performed of the changes in image quality and patient dose as a result of increasing filtration for fluoroscopy performed under automatic brightness control. Filtration was added either at the x-ray tube housing i.e., scatter-free geometry or adjacent to a tissue equivalent phantom simulating the patient i.e., with-scatter geometry. Patient doses were expressed in terms of the total energy imparted to patients simulated by either a 10 cm i.e., pediatric or 20 cm i.e., adult acrylic phantoms. Changes in image quality were determined by measuring the relative visibility of circular disks in a Leeds Test Object 10 contrast-detail phantom. In the scatter-free geometry, the addition of 4 mm Al filtration reduced the energy imparted by 27% 10 cm phantom and 20% 20 cm phantom. In the with-scatter geometry, the corresponding reductions in energy imparted were 17% and 9% for the 10 and 20 cm phantoms, respectively. The visibility of low contrast disks generally decreased as the thickness of the added aluminum increased but the location of the added Al i.e., with-scatter or scatter-free geometry had no significant effect on the resultant image quality. These results demonstrate that the use of patient support pads with a thickness of 4 mm Al will generally have an adverse impact on fluoroscopic image quality and result in modest reductions 10% of adult patient doses. 1997 American Association of Physicists in Medicine. S0094-2405 97 01403-X Key words: fluoroscopy, patient dose, image quality I. INTRODUCTION In diagnostic radiology, an increase in x-ray tube voltage and/or x-ray beam filtration will normally affect patient dose as well as the corresponding image quality. A higher x-ray beam quality, specified as a half-value layer and expressed in mm aluminum, results in a more penetrating beam. The exposure level at the image receptor is normally kept constant to maintain a fixed film density or image intensifier output signal. Any increase in x-ray beam quality will thus result in a reduction to the entrance skin dose and energy imparted to the patient. 1,2 Subject and image contrast are also reduced with increasing beam filtration and/or x-ray tube voltage for most radiographic and fluoroscopic x-ray examinations. 3,4 As a result, changing the x-ray beam quality requires a direct trade off between radiation dose and image quality in diagnostic radiology. Optimization of filtration and x-ray tube potential for a given type of radiographic examination needs to take both these factors into account. 5,6 Surgical procedures, which may include the use of fluoroscopy, can be extensive in time and often results in significant pressure sores. 7,8 To overcome this problem, gel pads may be used since they have been shown to significantly improve patient comfort. Gel pads with a thickness greater than 4 mm Al equivalent, measured at 100 kvp, have been reported. 9 Use of these patient support pads in fluoroscopy can increase both x-ray beam filtration as well as the x-ray tube kvp when used in automatic brightness control mode of operation. These patient support pads will therefore be expected to result in an increase in x-ray beam quality. Unlike conventional added x-ray beam filtration, which is added at the x-ray tube housing, most of the forward scattered radiation arising from interactions in the support pads will impinge on the patient. In addition, scattered radiation may also reach the image receptor which could degrade the fluoroscopic image quality. In this study, we investigated the effect of added filtration in fluoroscopy performed under automatic brightness control mode of operation. Filtration was added either at the x-ray tube housing scatter-free geometry or adjacent to a tissue equivalent phantom simulating the patient with-scatter geometry. The effect of adding filtration in each one of these two locations on the values of relative patient dose and image quality was investigated. Patient dose was expressed in terms of the energy imparted to the patient, simulated by either a 10 cm i.e., pediatric or 20 cm i.e., adult thick acrylic phantom. 10,11 Changes in image quality were determined by measuring the relative visibility of disks in a commercial contrast-detail phantom, the Leeds Test Object LTO 10. 12 II. METHOD A. X-ray equipment A Siemens Polystar fluoroscopy unit Siemens Medical Systems, Iselin, NJ, operated in automatic brightness control, was used in this experimental work. The x-ray unit employed a high frequency voltage generator, a Megalix 125/ 40/82 x-ray tube, and an Optilux 40-4 HD image intensifier. The video chain consisted of a Videomed H1X Saticon TV pickup tube and the image was displayed on a Sinomed 54 cm monitor. Figure 1 shows the kvp vs ma fluoroscopy curve used by this imaging system as supplied by the manufacturer. Phantom material for simulating a patient consisted 377 Med. Phys. 24 (3), March 1997 0094-2405/97/24(3)/377/6/$10.00 1997 Am. Assoc. Phys. Med. 377

378 Geiser, Huda, and Gkanatsios: Effect of patient support pads on image quality 378 of contrast when the test phantom is placed in the x-ray field and viewed on a monitor. 12 Image quality was taken to be the total number of disks visible in the displayed image of the LTO 10 phantom during fluoroscopy. Figure 2 shows the location of the LTO 10 phantom attached to the center of the image intensifier which was used in the 20-cm-diam mode. The number of disks seen in the images of the LTO 10 phantom in these experiments ranged from 30% to 55% of the total number 108 actually present. Five readers recorded the total number of disks visible in each group of the LTO 10 phantom. Viewing was done during live fluoroscopy and images were displayed in a darkened room on an optimized display monitor. Readers were allowed to allocate a score of 0.5 to poorly visualized disks. Two readers also made ten additional observations each of disk visibility in the LTO 10 phantom image with 4-mmthick Al filter. The location of the added filter was randomly varied between the scatter-free and with-scatter positions see Fig. 2. FIG. 1. Fluoroscopy ma vs kvp curve used by the Polystar operating in automatic brightness control mode. of 2.5-cm-thick slabs of Lucite methyl methylacrylic with a density of 1.19 g/cm. Phantom thicknesses of 10 and 20 cm were used to simulate pediatric and adult patients, respectively. Figure 2 shows a schematic of the experimental setup for a 20-cm-thick patient. Measurements were made with up to 6 mm of added aluminum filtration located either at the x-ray tube collimator scatter-free geometry or the entrance of the phantom with-scatter geometry. B. Image quality Relative changes in image quality were measured using a LTO 10 phantom, containing 12 groups of objects ranging from 0.75 to 11 mm in diameter. The circular disks are made of different materials and thicknesses to give different levels FIG. 2. Schematic diagram of the experimental arrangement used to determine image quality and radiation doses. C. Radiation dosimetry The energy imparted to a water phantom,, can be expressed as p X p s X s A J, 1 where is a parameter which converts the area-exposure product into energy imparted to the phantom and which depends on the x-ray kvp, beam quality HVL and the phantom thickness, X is the exposure (R) at the phantom surface measured without the phantom being present i.e., free-inair, and A is the corresponding x-ray beam cross-sectional area cm 2. The subscripts p and s refer to the primary and scatter components of a filtered x-ray beam. For the scatterfree geometry, only the primary beam contributes to the energy imparted to the phantom. Use of the with-scatter geometry results in an additional contribution from scattered radiation as given by the term X s in Eq. 1. The parameter is given by HVL Jcm 2 R 1, 2 where HVL is the measure half-value layer for the appropriate primary or scatter component of the x-ray beam and and are parameters which depend on the x-ray beam kvp and phantom thickness. 13,14 Figure 3 shows the computed values of the parameter for the primary Polystar x-ray output as a function of applied kvp for 10 and 20 cm water phantoms. No corrections were made to the data shown in Fig. 3 to account for the different densities of water and acrylic as we are only interested in the relative values of energy imparted to the simulated patient. In general, increases with increasing kvp, HVL, and phantom thickness. During fluoroscopy with a given combination of phantom thickness and added Al filtration, the x-ray technique factors selected by the Polystar were locked in place after the system stabilized. Subsequent measurements of exposure and x-ray beam HVL were done using these technique factors, but with the Polystar unit operated in manual mode. The primary

379 Geiser, Huda, and Gkanatsios: Effect of patient support pads on image quality 379 FIG. 3. Energy imparted to a water phantom per unit exposure-area product i.e., vs x-ray tube kvp. beam exposure (X p ) at the phantom entrance was determined in the scatter-free geometry using a MDH 1015 Radcal Corporation, Monrovia, CA exposure meter with a 10 5 6 chamber. Moving the Al filter adjacent to the ionization chamber resulted in a measured exposure which included primary and scatter i.e., X p X s. The entrance skin exposure due to the scatter radiation from the Al filter (X s ) was obtained by subtracting the primary exposure from the corresponding primary plus scatter exposure measurement. Five separate measurements of each exposure value were taken and the corresponding mean value and standard deviation were computed. The scatter component was studied theoretically by employing x-ray spectra calculated using the model developed by Tucker et al. 15 The number of scattered photons produced by an aluminum filter with thickness t mm was determined using the Compton scattering theory of Klein and Nishina. 16 The small contributions of coherent scatter and multiple scatter in the Al were neglected. For each incident photon energy, the number and corresponding energy of the scattered photons at each angle were obtained using the Compton scatter cross section data of Evans 16 and computed at 1 intervals between 0 and 90. 17 The total scatter was obtained by summing the contributions from all the photons in the primary x-ray spectrum. The HVL of the primary photon beam, corresponding to X p, was measured directly for the scatter-free geometry. The half-value layer of the computed scatter spectra were obtained using an iterative procedure to find the thickness of Al which reduced the exposure from the x-ray spectrum by 50%. 13 Figure 4 shows the calculated HVL values for both the primary and scatter components of an x-ray beam filtered by 2, 4, and 6 mm added Al filtration where the initial x-ray spectrum was generated by a tungsten target 10% rhenium atoms, a 12 anode angle, and 3 mm inherent aluminum filtration. FIG. 4. Calculated HVL vs kvp for the primary and scattered x-ray components as transmitted through the specified amounts of added aluminum filtration. III. RESULTS The technique factors selected by the automatic brightness control system during fluoroscopy are listed in Table I. TABLE I. Technique factors set by the Polystar automatic brightness control system during fluoroscopy. Added aluminum mm 10 cm phantom kvp/ma 20 cm phantom kvp/ma 0 61/1.0 80/2.6 2 63/1.1 82/2.8 4 65/1.2 84/3.0 6 67/1.3 86/3.2

380 Geiser, Huda, and Gkanatsios: Effect of patient support pads on image quality 380 TABLE II. Total score of visible disks in the LTO 10 phantom image for each of five readers. Added aluminum mm Reader LTO 10 image score for the 10 cm/20 cm phantom #1 #2 #3 #4 #5 Average score 0 62/42 63.5/39.5 59/34 52.5/34 59/32 59.2/36.3 2 59/38.5 62.5/38.5 58/35.5 53.5/35.5 52/31 57/35.8 4 58.5/35 63/32 54/31.5 54.5/31.5 51.5/31 56.3/32.2 a 6 53.5/34 52/34 51/31 48.5/31 47.5/29 50.5 a /31.8 a a P 0.01. The corresponding low contrast detection results for each of five observers for the LTO 10 phantom images are given in Table II. The average number of visible disks decreased with increasing thickness of the added Al. Also shown are the statistically significant results of a paired t-test analysis performed using the differences between the number of disks seen with no added Al and the number seen with the specified amount added Al. Table III summarizes the results obtained for two readers who read the images of the LTO 10 phantom with 4 mm Al randomly positioned in the scatterfree and with-scatter locations. The differences between the two filter locations shown in Table III were not statistically significant. Table IV lists the measured HVLs and free-in-air entrance exposure rates for both the scatter-free and with-scatter geometries. Figure 5 shows the corresponding values of energy imparted to each phantom as a function of added Al filtration for both with-scatter and scatter-free geometries. The error bars shown in Fig. 5 correspond to the computed standard errors associated with five repeat readings. For the scatterfree geometry, the addition of 4 mm Al filtration reduced the energy imparted by 27% 10 cm phantom and 20% 20 cm phantom. For the with-scatter geometry, the corresponding reductions in energy imparted were 17% and 9% for the 10 and 20 cm phantoms, respectively. IV. DISCUSSION The / parameters used in Eq. 2 were derived for use with conventional clinical x-ray spectra and give rise to parameter values with uncertainties of 1%. 14 Use of this equation for the scatter component was empirically validated by comparing the values of obtained using Eq. 2 for scatter spectra with those obtained using analytical values for monoenergetic photons. 13,17 The average difference in the TABLE III. Scores of visible disks in the LTO 10 phantom image with 4 mm added filtration. Each score is the mean of five readings. 10-cm-thick phantom. Reader Scatter-free With-scatter 1 53.3 0.6 53.4 1.6 2 53.1 0.7 53.8 1.4 values for these two methods for three scatter spectra at x-ray tube voltages of 60, 80, and 100 kvp, and obtained with a 2 mm aluminum filter as the scattering material, was 0.2%. This excellent agreement demonstrates the validity of employing Eq. 2 for use with the scatter spectra investigated in this study. During fluoroscopy the automatic brightness control mechanism maintained a constant entrance exposure to the image intensifier. As the thickness of the added aluminum was increased, the Polystar increased the kvp and the ma along the curve depicted in Fig. 1 in order to maintain a constant exposure level at the image intensifier. Modern fluoroscopy systems such as the Polystar have a series of operator selected curves, analogous to the one shown in Fig. 1, which permit either the image quality to be improved or the patient dose to be reduced. A vertical kvp vs ma curve, for example, would result in reduced patient dose at the price of reduced image quality whereas a horizontal curve would maintain image quality at the price of increased patient dose. The results obtained in this study pertain to the kvp vs ma curve used at our institution for adult patients. This kvp vs ma curve is an intermediate one between the two extremes which would solely focus on either the image quality or the patient dose. The precision of the subjective image quality measurements are given in Table III with an average variability in the number of disks observed being about 1.1 where the total number of disks observed was 53. The scores for the five readers listed in Table II, however, show larger absolute differences in disk visibility. This inter-reader variability is a well-known feature of image quality assessments obtained using contrast-detail phantoms and results from the adoption of variable decision thresholds by different readers. 18 The uncertainties associated with the exposure measurements are mainly due to errors in positioning the ion chamber and small fluctuations in fixing the x-ray technique factors. The average uncertainty in the entrance skin exposure for a simulated patient was 5%. TABLE IV. Summary of measured half-value layers and exposure rates. Added filtration mm Al Primary HVL mm Al 10 cm/20 cm X p a R/min 10 cm/20 cm X p X s b R/min 10 cm/20 cm 0 2.2/2.9 0.54 0.03/2.41 0.13 0.54 0.03/2.41 0.10 2 3.2/4.2 0.34 0.02/1.64 0.07 0.38 0.02/1.80 0.07 4 3.8/5.0 0.27 0.02/1.29 0.06 0.32 0.02/1.51 0.06 6 4.4/5.7 0.21 0.01/1.05 0.06 0.26 0.01/1.31 0.06 a Scatter-free geometry. b With-scatter geometry.

381 Geiser, Huda, and Gkanatsios: Effect of patient support pads on image quality 381 FIG. 5. Energy imparted to an acrylic phantom vs added Al filtration for an x-ray beam with an area of 154 cm 2 at the phantom surface 1 min exposure. Decreases in image quality with added Al filtration as depicted in Table II are due to the increase in kvp and beam hardening. The automatic brightness control system increased the x-ray beam kvp to maintain a constant input exposure to the image intensifier as the aluminum thickness was increased. In addition, the aluminum placed in the x-ray beam hardens the x-ray beam. Both of these factors serve to increase the effective energy of the x-ray beam which results in a reduced subject contrast. As with most contrast-detail phantom studies, it is difficult to relate the loss of image quality for phantom studies as depicted in Table II to clinical studies. The importance of any loss of image quality will be generally task dependent and will likely affect the detection of subtle low contrast features to a greater extent than for high contrast features containing barium or iodine contrast. Having the aluminum closer to the image intensifier could degrade the image quality if scatter from the aluminum reached the input surface of the image intensifier. Our experimental data, however, showed that the location of the added Al filtration did not have any significant effect on the resultant image quality. This result is not unexpected since there is a relatively large distance from the phantom surface to the image intensifier which will act as an air gap. This minimizes the number of scattered photons which could reach the image receptor. In addition, the image intensifier has a 17:1 grid 70 lines/cm which would also serve to minimize the number of scattered photons reaching the image intensifier. The relative increase in exposure i.e., X s /X p due to scatter by the introduction of 2 6 mm Al was in the range of 0.1 0.25. As expected, the HVL of the scatter radiation was generally lower than that of the primary photon beam. With 2 mm Al filtration, the HVL of the scattered radiation was 15% lower than that of the transmitted primary radiation and with 6 mm Al filtration, the scattered radiation HVL reduced by 25%. At 80 kvp, a reduction of the x-ray beam HVL from 3 to 2.4 mm i.e., 20% would correspond to a reduction in the parameter by 12% and 14% for 10 and 20 cm water phantoms, respectively. Energy imparted is better than the entrance skin exposure as an indicator of the patient dose and corresponding patient risk. 19 In this study, the dose savings associated with increasing filtration were much lower when expressed in terms of energy imparted than with the entrance skin exposure parameter. For example, adding 4 mm Al filtration in the scatterfree geometry reduces the entrance skin exposure for a 10 cm acrylic phantom by 50% but the corresponding reduction in energy imparted is only 27%. This finding is in very good agreement with published data and results from the fact that the entrance skin exposure fails to take into account the quality of the incident x-ray beam. 5,6 The results of Fig. 5 show a monotonic decrease in the energy imparted as the aluminum thickness was increased. A linear curve fit show that in the scatter-free geometry, dose savings for adults were 4.6%/mm added Al. In the case of with-scatter geometry, the corresponding dose savings were only 2.2%/mm added Al. For adult patients, the dose savings achieved using the with-scatter geometry are therefore about a factor of 2 lower than those obtained from using a scatterfree geometry. The results in this study indicate that a 4 mm Al equivalent filter pad will have an adverse impact on image quality and likely reduce adult patient doses by 10%. Clinical use of such pads should therefore require the patient benefits in terms of improved comfort, reduced risk of sores, and reduced patient dose to be weighted against the corresponding loss of image quality. In the absence of any knowledge of the specific imaging tasks required during patient fluoroscopy, general recommendations are clearly not possible. Nevertheless, the results obtained in this study show that patient dose savings are generally modest and unlikely to be a significant factor in any decision about the benefits to be achieved by the use of thick patient support pads. The use of these support pads will most likely be justified by arguing that any loss in image quality would be outweighed by the improved patient comfort and reduced risk of incurring pressure sores. ACKNOWLEDGMENTS The authors are grateful to Lynn Rill, Angela Properzio- Bruner, and Guoying Qu for assistance in scoring the images

382 Geiser, Huda, and Gkanatsios: Effect of patient support pads on image quality 382 of the LTO 10 phantom. We would also like to thank Cecile Mohr, Angela Properzio-Bruner, and Lynn Rill for their helpful discussion and comments. a Presented at the 1996 AAPM meeting in Philadelphia Med. Phys. 23, 1059 1996. b Corresponding author: Phone 352 395-0293; Fax 352 395-0279; Electronic mail hudaw@xray ufl.edu 1 D. Trout, J. Kelley, and G. Cathey, The use of filters to control radiation exposure to the patient in diagnostic roentgenology, Am. J. Roentgenol. 62, 946 963 1952. 2 P. Shrimpton, D. Jones, and B. Wall, The influence of tube filtration and potential on patient dose during x-ray examinations, Phys. Med. Biol. 33, 1205 1212 1988. 3 P. Sprawls, X-ray imaging concepts. Basic considerations, AAPM Monogr. No. 20, 41 74 1994. 4 M. Tapiovaara and M. Sandborg, Evaluation of image quality in fluoroscopy by measurements and Monte Carlo calculations, Phys. Med. Biol. 40, 589 607 1995. 5 K. Koedooder and H. Venema, Filter materials for dose reduction in screen-film radiography, Phys. Med. Biol. 31, 585 600 1986. 6 H. Nagel, Comparison of performance characteristics of conventional and K-edge filters in general diagnostic radiology, Phys. Med. Biol. 34, 1269 1287 1989. 7 F. Gendron, Burns occurring during lengthy surgical procedures, J. Clin. Eng. 5, 19 26 1980. 8 S. Stewart, V. Plamieri, and V. Cochran, Wheelchair cushion effect on skin temperature, heat flux, and relative humidity, Arch. Phys. Med. Rehabil. 61, 229 233 1980. 9 W. Huda, S. Knapp, and M. Barrett, Influence of table support pad on patient dose and image quality in mobile C-arm fluoroscopy, Radiology P185, 304 1992. 10 J. Persliden and G. Carlsson, Energy imparted to water slabs by photons in the energy range 5 300 kev. Calculations using a Monte Carlo photon transport model, Phys. Med. Biol. 29, 1075 1088 1984. 11 P. Shrimpton, B. Wall, D. Jones, and E. Fisher, The measurement of energy imparted to patients during diagnostic examinations using the Diamentor exposure-area product meter, Phys. Med. Biol. 29, 1199 1208 1984. 12 G. Hay, O. Clarke, N. Coleman, and A. Cowen, A set of x-ray test objects for quality control in television fluoroscopy, Br. J. Radiol. 58, 335 344 1985. 13 N. Gkanatsios, Computation of energy imparted in diagnostic radiology, MS thesis, University of Florida, 1995. 14 N. Gkanatsios and W. Huda, Computation of energy imparted in diagnostic radiology, Med. Phys. in press, 1997. 15 M. Tucker, G. Barnes, and D. Chakraborty, Semi-emperical model for generating tungsten target x-ray spectra, Med. Phys. 18, 211 218 1991. 16 R. D. Evans, The Atomic Nucleus Krieger, Malabar, FL, 1955. 17 W. Geiser, Effects of patient support pads on image quality and dose in fluoroscopy, MS thesis, University of Florida, 1996. 18 International Commission on Radiological Units and Measurements ICRU Report No. 54, Medical imaging The assessment of image quality, Bethesda, MD, 1995. 19 B. Wall, R. Harrison, and F. Spiers, Patient dosimetry techniques in diagnostic radiology, Report No. 53 Institute of Physics Sciences in Medicine IPSM, York, England, 1988, pp. 47 55.