A comparative study of several digital flat panel X-ray units: patients doses and image quality in chest radiography Torres Cabrera R. 1, España López M.L. 2 Ruiz Manzano P. 3, Sastre Aguado J.M. 4,, Rivas Ballarin M. A. 3, Ferrer García N. 4, Hernando González I 1.,Peinado Montes M. 1, Floriano Pardal A. 2, Arranz y Carrillo de Albornoz L. 4, López Franco P. 2 1 Servicio de Radiofísica y Protección Radiológica. H. U. Río Hortega Valladolid, Spain E-mail: rtorres@hurh.sacyl.es 2 Servicio de Física y Protección Radiológica. H. U. La Princesa, Madrid, Spain E-mail: mespana.hlpr@salud.madrid.org 3 Servicio de Física y Protección Radiológica. H.C.U. "Lozano Blesa", Zaragoza, Spain E-mail: fpro-pr@hcu-lblesa.es 4 Servicio de Física y Protección Radiológica. H.U. Ramón y Cajal, Madrid, Spain E-mail: jsastrea.hrc@salud.madrid.org Abstract. The aim of this study is to evaluate and compare image quality and patients entrance surface dose (ESD) in PA and LAT chest examinations given by several digital flat panel X-ray units. Four hospitals are involved: H. C. U. Lozano Blesa of Zaragoza, H. U. de la Princesa of Madrid, H. U. Ramón y Cajal of Madrid, and H. Clínico Universitario of Valladolid. ESD received by 50 standard patients has been estimated, both for PA and LAT projections, by using the tube output and the radiographic technique (selected kvp, mas, patient thickness and focus-to-detector distance), assuming a standard backscattering factor (b f = 1.35). Average values for ESD in PA chest and in LAT chest are much lower than reference values (PA:0,3 mgy; LAT: 1,5 mgy) although a wider range can be seen in the mean values, partly due to the use of additional filtration. Image quality has been evaluated with an anthropomorphic phantom (QC Phantom for digital and conventional chest radiography, Nuclear Associates 07-646) and a contrast-detail test pattern (CDRAD type 2.0, Instrumentele Dienst). There is a direct relation between dose and image quality, so optimization processes should be carried out in order to ensure a minimum dosis compatible with an image quality suitable for diagnosis. 1. Introduction Direct capturing of digital images in radiology (DR) by means of flat-panel X-ray image detectors is being progressively implemented in medical practice due to the multiple benefits of this technique: its capacity for providing higher-quality images, the possibility of image processing, the dramatic reduction of rejection rate, and a much longer range of radiation response than any given screen/film conventional system. Additionally, these systems possess a higher DQE (Detective Quantum Efficiency), making it possible to obtain images with the same noise level as in a conventional system but with lower radiation doses [1]. The relationship between radiation dose and image quality on different digital equipment, upon the analysis of phantom [2, 3] animal or [4] images, has been looked into in different publications. Some authors [5] recommend to add filtration for chest examinations in digital equipment, with the aim of reducing the dose as well as achieving an image quality similar to the one obtained with a conventional spectrum. The implementation of this new technology requires an estimation of the doses that are actually being administered in clinical practice, in order to check that, in cases of both day-to-day practice and optimisation protocols, doses are kept within reference values [6, 7] and as low as achievable in relation to the aimed image quality. The aim of this study is to evaluate and compare image quality and patients Entrance Surface Dose (ESD) in chest examinations (two projections PA and LAT) in flat-panel digital equipment in four 1
reference hospitals H. C. U. Lozano Blesa of Zaragoza, H. U. de la Princesa of Madrid, H. U. Ramón y Cajal of Madrid, and H. Clínico Universitario of Valladolid. 2. Materials and Methods The study has been carried out with four digital equipments used for chest examinations, whose characteristics are presented in Table I. Generator Tube Additional filtration Table I. Digital systems characteristics H1 H2 H3 H4 SIEMENS POLYDOROS LX 50 LITE SIEMENS OPTILIX 150/30/50C 0,1 / 0,2 / 0,3 mm Cu TOSHIBA KX0-60G TOSHIBA DRX- 3724 HD --- PHILIPS OPTIMUS 50 PHILIPS RO 1750 0,1 mm Cu 1 mm Al GE Advant X-E GE Maxi Ray 100 --- Detector Trixell CANON CXDI- 40G Trixell GE Revolution TM Size 43 x 43 cm 43 x 43 cm 43 x 43 cm 41 x 41 cm Matrix (pixels) 3000 x 3000 2688 x 2688 3000 x 3000 2048 x 2048 Scintillator plate CsI Gd 2 O 2 S:Tb CsI CsI Anti scatter grid R:15/ N:80 f:180 R:12/ N:44 f:180 R:12/ N:36 f:180 R:13/ N:78 f:180 Sensitivity class 400 400 400 400 Hard copy system FUJI FM-DP L AGFA Drystar 4500 FUJI Drypix 7000 KODAK Ektascan 2180 All instruments are programmed for an equivalent class sensitivity of 400, therefore the detector entrance dose should be approximately equivalent to 2,5 µgy per image, and they undergo regular quality controls [6] to check that all physical parameters are within tolerance values. Expected overall filter values and system performance values are shown in table II, including in the case of H1 both X- ray spectrum qualities used, without added filtration (H1) and with a 0,1 mm Cu filter (H1.1) Table II. Estimated Total Filtration (mm Al) and tube output at 1m( µgy/mas a 1 m.). Total Filtration 3,2 6,6 3,42 7,86 2,85 Tube output (80 kvp) 62,3 28 49,1 25,6 61 Tube output (kvp set) 132,8 89,9 122 74,7 124 Multimeter PMX-III PMX-III Victoreen 4000 M PMX-III Detector R-25 R-25 R-25 The dosimetric index value used has been the Entrance Surface Dose (ESD) for both chest projections (PA and LAT), with a sample of 50 patients in each facility. In centre H1, samples have been taken with and without an additional filtration of 0,1 mm Cu and in centre H3 all examinations have been 2
carried out with an additional filtration of 0,1 mm Cu + 1 mm Al. ESD has been estimated for each patient in the sample on the basis of actual parameters used in the examination and the previously measured tube output, assuming a standard backscattering factor (b f = 1.35). Image quality has been evaluated with an anthropomorphic phantom (QC Phantom for digital and conventional chest radiography, Nuclear Associates 07-646) and a contrast-detail test pattern (CDRAD type 2.0, Instrumentele Dienst). Images have been evaluated by four independent experts separately, in the same format used by the radiologists of each centre, which in most cases has been hard copy system. 3. Results and discussion Patients doses Mean values of kv, mas and thickness under aforementioned conditions are shown in tables III and IV. Table III. Mean and standard deviation values. Chest PA projection mas (SD) 1,05 (0,19) 1,25 (0,19) 1,53 (0,82) 0,73 (0,23) 1,84 (0,34) Thickness (SD) (cm) 24 (1,8) 24,3 (1,8) 25,9 (2,8) 25,5 (2,2) 24,9 (1,7) kvp set (SD) 125 (0) 125 (0) 130 (0) 125 (0) 125 (1.1) Table IV. Mean and standard deviation values. Chest LAT projection mas (SD) 3,4 (1) 3,83 (1) 4,87 (2,33) 2,9 (1,12) 7,15 (2,9) Thickness (SD) (cm) 34 (2,3) 33,4 (2,4) 33,5 (2,8) 34,5 (2,6) 33,7 (2,3) kvp set (SD) 125 (0) 125 (0) 130 (0) 125 (0) 125 (1.1) Patients samples have got very similar thickness in the four centres, so we can consider them standard patients. Mean values of ESD are shown in tables V and VI. These values are much lower than reference values (PA:0,3 mgy; LAT: 1,5 mgy). ESD (SD) (mgy) Table V. Mean values and standard deviation. Chest PA projection FLAT PANEL SYSTEM 0,08 (0,015) 0,065 (0,01) 0,11 (0,06) 0,03 (0,01) 0,13 (0,025) Table VI. Mean values and standard deviation. Chest LAT projection FLAT PANEL SYSTEM ESD (SD) (mgy) 0,3 (0,093) 0,23 (0,064) 0,38 (0,17) 0,14 (0,05) 0,56 (0,24) The range of mean ESD values for the different systems is very wide, both for chest PA (0,03-0,13 mgy) and for chest LAT (0,14-0,56 mgy). This could be a consequence of the use of additional filters in hospitals 1 and 3. The consequence of adding 0,1 mm Cu is clearly shown in Hospital 1, where its use causes almost a 20% reduction in ESD, although this dosimetric index, useful and easy 3
to measure, may be not suitable enough when comparing doses between systems whose radiation spectra are quite different. In those cases other magnitudes that correlate better with the effective dose should be considered. Image quality FIG. 1. Evaluation of low-contrast systems performance FIG. 2. Evaluation of low-contrast systems performance corresponding to an exposure of 10 µgy at the wall-stand device entrance The results obtained with a CDRAD phantom are shown in figure 1. In each case, the radiographic technique and processing algorithm for a PA chest projection were applied. There is a significant difference regarding image quality between hospital H4 (which also applies the highest doses) and the other centres. The results may be even more relevant when image quality for the same entrance dose to the wall-stand device (10 µgy) is compared, as shown in figure 2. In this case, the number of objects detected in hospitals H3 and H4 is very similar, and superior to those detected in hospitals H1 and H2. It seems clear that increasing the dose in H3, image quality improves remarkably. Another result to be underlined is the difference in image quality between hospitals H1 and H3, although both include the same digital detector. Nevertheless there are other parameters which have got an effect on image quality (anti-scatter grid, processing algorithm used, etc.). Among all these factors, the image post- processing carried out by each system is particularly important, and that could give rise to significant differences. Figures 3 and 4 show the results for the chest PA technique obtained with an anthropomorphic phantom. Again, the highest number of detected objects corresponds to hospital H4 that, as can be seen in the curve corresponding to optic densities, presents a higher contrast in the lung area. If we compare figure 3 with figure 5, obtained with a dose of 10 µgy at the entrance of the wall-stand device, we can appreciate a remarkable improvement on image quality with increasing the dose. Thus, it could be inferred that in digital radiology there is a direct relation between dose and image quality. Therefore, optimisation processes should be carried out in order to ensure a minimum dose compatible with an image quality suitable for diagnosis. 4
The anthropomorphic phantom also includes a line-pair test pattern to evaluate spatial resolution. Results are consistent with pixel size in each detector, which determines the limiting frequency or Nyquist frequency: 2,5 pl/mm in hospital H4, 3,1 pl/mm in hospital H2, and 3,5 pl/mm in hospitals H1 and H3. In hospital H1 and H3, as their detectors consist of four assembled smaller panels, both the images obtained with the anthropomorphic phantom and those obtained with a resolution uniformity grid show a blind line at the interface. This transition zone is more visible in hospital H1 than in H3. In addition, correct visualisation of the resolution uniformity grid becomes particularly difficult in hospital H2, which is consistent with the fact that its scintillator is Gd 2 O 2 S:Tb, which, in principle, should produce a higher light diffusion than in the other cases, where CsI was used. FIG. 3. Test objects detected in the antropomorphic phantom. Chest PA radiographic technique and automatic control exposure 4. Conclusions The Entrance Surface Dose, measured in patients undergoing chest examinations in the digital equipments evaluated, does not exceed the reference dose values, although a wider range can be seen in the mean values of this dosimetric index, partly due to the use of additional filtration. Regarding image quality, bearing in mind that in all cases an image with diagnostic quality is obtained, in some centres quality could be improved by increasing the doses to levels similar to those of other centres. On the other hand, a dose reduction is often possible if a moderate decrease in image quality is accepted. Last, the monitor characteristics, or, in this case, the type of printer and its calibration, are not less important. In order to avoid their influence on the comparison of image quality, it would be necessary to capture images in digital format, visualising or printing them in the same device, but this is still not possible in a simple way in some of the equipment evaluated. FIG. 4. Optical densities measured in several thoracic regions FIG. 5. Test objects detected in the antropomorphic phantom corresponding to an exposure of 10µGy at wall-stand unit entrance 5
References 1. Harrell G. Chotas MS, James T. Dobbins, III, PhD and Carl E. Ravin MD. Principles of digital radiography with large-area, electronically readable detectors: a review of the basics Radiology 1999; 210: 595-599 2. Hamer OW, Volk M, Zorger N, Feuerbach S, Strotzer M. Amorphous Silicon, Flat-Panel, X-Ray Detector Versus Storage Phosphor-Based Computed Radiography: Contrast-Detail Phantom Study at Different Tube Voltages and Detector Entrance Doses Invest Radiol 2003 Apr;38(4):212-220. 3. Marshall NW, Faulkner K, Busch HP, Marsh DM, Pfenning H. A comparison of radiation dose in examination of the abdomen using different radiological imaging techniques Br J Radiol 1994 May;67(797):478-84. 4. Karl Ludwig MD, Kathrin Ahlers BS, Dag Wormanns MD et als. Lumbar spine radiography: Digital Flat Panel detector versus screen-film and storage-phosphor systems in Monkeys as a pediatric model Radiology 2003; 229:140-144. 5. Dobbins JT, Samei E, Chotas HG, Warp RJ et als Chest radiography: Optimization of X-ray spectrum for cesium iodide-amorphous silicon flat panel detector Radiology 2003; 226:221-230. 6. Protocolo español de control de calidad en radiodiagnóstico. (Revisión 1). Aspectos técnicos. SEFM-SEPR 2003. 7. European guidelines on quality criteria for diagnostic radiographic images. EUR 16260 EN. Junio 1996. Key words: X-ray dosimetry, digital, radiodiagnosis, flat panel, image quality 6