Studies on reduction of exposure dose using digital scattered X-ray removal processing Poster No.: C-1834 Congress: ECR 2015 Type: Scientific Exhibit Authors: K. Kashiyama, M. Funahashi, T. Nakaoka, T. Nakamura, N. 1 2 1 3 1 3 1 3 3 Amimoto, K. Okano, M. Yamada, T. Kawamura, S. Naito ; 1 2 3 Osaka/JP, Tokyo/JP, Kanagawa/JP Keywords: Computer applications, Radioprotection / Radiation dose, Anatomy, Digital radiography, Computer Applications-Detection, diagnosis, Computer Applications-General, Radiation effects, Workforce DOI: 10.1594/ecr2015/C-1834 Any information contained in this pdf file is automatically generated from digital material submitted to EPOS by third parties in the form of scientific presentations. References to any names, marks, products, or services of third parties or hypertext links to thirdparty sites or information are provided solely as a convenience to you and do not in any way constitute or imply ECR's endorsement, sponsorship or recommendation of the third party, information, product or service. ECR is not responsible for the content of these pages and does not make any representations regarding the content or accuracy of material in this file. As per copyright regulations, any unauthorised use of the material or parts thereof as well as commercial reproduction or multiple distribution by any traditional or electronically based reproduction/publication method ist strictly prohibited. You agree to defend, indemnify, and hold ECR harmless from and against any and all claims, damages, costs, and expenses, including attorneys' fees, arising from or related to your use of these pages. Please note: Links to movies, ppt slideshows and any other multimedia files are not available in the pdf version of presentations. www.myesr.org Page 1 of 25
Aims and objectives In Japan, x-ray examination of the chest region is carried out with portable equipment on medical wards for the management of the critically-ill patient. These patients whose condition is so severe that they cannot be transferred to an x-ray room, have a portable chest x-ray examination from which the diagnostic information is vital in the patient's treatment. Generally, when performing the portable chest x-ray examination the thicker the subject, the more scattered the x-ray beam will be in the x-ray exposure and the greater its effect on image quality. Some departments take the approach of reducing the kvp used for the portable chest examination thereby reducing the scatter radiation that reaches the image receptor. However, this also reduces the energy of the x-ray photons and the penetration of structures in the chest, resulting in reduced diagnostic information. In some cases, a secondary radiation grid is used with the higher kvp. The higher kvp maintains the penetration of structures in the chest region and the grid reduces the influence of the scattered x-rays, thereby improving image quality and diagnostic information. However, using a grid in a portable environment presents workflow challenges such as carrying the grid, positioning the grid and FPD perpendicular to the x-ray beam. It also creates its own image quality issues in the way of artefacts from the misalignment of grid and x-ray beam (often referred to as "grid cut-off") which can also result in reduced diagnostic information. Due to these reasons, many hospitals do not use a grid in portable chest examinations performed on wards, opting to reduce the kvp used and accept the compromise between image quality (diagnostic value) and acceptable practical workflow. The Osaka General Medical Center currently uses CR for all portable chest x-ray examinations. No grid is used for portable chest x-ray examinations, but a low tube voltage of 65kVp is used to reduce the effect of scattered x-rays. In conjunction with the newly introduced flat panel detector (FPD) made by Fujifilm for portable examination at Osaka General Medical Center a study was undertaken looking at how newly developed scatter removal processing (Virtual Grid/VG) could be used to improve image quality in portable chest examinations with an associated dose reduction when compared to using a standard secondary radiation grid (+). The results of this study are presented here. Scattered x-ray causes a reduction of contrast and increased granularity in an image. Virtual Grid processing has [Contrast Improvement Processing] and [Granularity Improvement Processing] to improve both contrast and granularity (Fig.1). Contrast Improvement Processing estimates scattered x-ray from the input image and reduces the Page 2 of 25
scattered x-ray in a ratio similar to a normal secondary radiation grid to improve image contrast. In addition, since scattered x-ray contains almost no information related to the subject structure, Granularity Improvement Processing extracts the granular components not related to the structure of the subject from the image and reduces the extracted components, improving structural definition in the image. An example image with Virtual Grid processing applied is shown in Fig.2. The left image is an image exposed without a grid (-), the middle image is an image exposed with a secondary radiation grid (+) with a grid ratio of 8:1. When these two images are compared, it is obvious that the left image has less contrast due to scattered x-ray. The right image is the same image as shown on the left but with Virtual Grid processing applied that corresponds to a grid ratio of 8:1 (as used in the middle image). As can be seen the contrast of the non-grid (-) image (left) has been improved to the level of that shown in the grid image (middle) by application of Virtual Grid processing. Images for this section: Fig. 1 Page 3 of 25
Fig. 2 Page 4 of 25
Methods and materials For the study, a Caesium Iodide FPD was used for image acquisition. Two Image sets were acquired, one set using a secondary radiation grid (+) exposure technique and the other set using a non-grid (-) exposure technique. The non-grid (-) images were then subjected to Virtual Grid processing. All image sets were evaluated using 3 methods. [1] Physical evaluation A contrast phantom (a 1cm-thick-PMMA block with a 2cm x 2cm square shaped hole) was placed between PMMA blocks with a thickness of 4cm and 4.5cm (9.5cm in total), and 7cm and 6.5cm (14.5cm in total). X-ray tube voltages of 60kVp and 80kVp, were used and a target exposure dose of 1mR at the back of the phantom was set as 100% (Table 1). Image sets were acquired with a grid (+) and without a grid (-) at 100%, 50% and 25% dose levels for both 60kVp and 80kVp. Virtual Grid processing corresponding to grid ratios of 6:1 was applied to the images acquired without a grid (-). The characteristics of contrast (C = D0#D10), the characteristics of noise (N = (SD0#SD10) / 2) and the contrastnoise ratio (CNR = C / N) were calculated setting the average pixel value of the 2cm x 2cm square shaped hole (0cm thick) of the contrast phantom and the area of the same size (1cm thick) next to the hole as D0 and D10, and the standard variation as SD0 and SD10. [2] Visual evaluation As shown in Fig.3, A CDRAD phantom was placed between PMMA blocks with thicknesses of 4cm and 4.5cm (9.5cm in total), and 7cm and 6.5cm (14.5cm in total) (Fig.3). X-ray tube voltages of 60kVp and 80kVp were used and a target exposure dose of 1mR at the back of the phantom was set as 100%. Image sets were acquired with a grid (+) and without a grid (-) at 100%, 50% and 25% dose levels (Table 1) for both 60kVp and 80kVp. (Table.1). Virtual Grid processing corresponding to grid ratio of 6:1 was applied to the images acquired without a grid (-). The CDRAD phantom images were analysed using CDRAD Analyser software ver. 2.1 (manufactured by Artinis Medical Systems). Contrast Detail curves were produced, and the IQFinv calculated for each image (Fig.4). The higher the IQFinv value, the greater the visibility of the details within the image. [3] Chest phantom evaluation Using a chest phantom, an x-ray tube voltage of 80kVp and the exposure dose of 1.8mAs, image sets were acquired with a 6:1 ratio grid (+) and without a Grid (-). Virtual Grid processing corresponding to a grid ratio of 6:1 was applied to the images acquired without a grid (-). Virtual Grid processed images and images exposed with a grid (+) were compared using the Kyoto University method* (one of the representative methods Page 5 of 25
in Japan used to assess the anatomical index and the physical factor of the chest region) below: Peripheral side of the lung field (resolution of peripheral blood vessel in S2 at upper right lung and resolution of the peripheral blood vessel of costophrenic region in S8 at bottom right lung). Mediastinum side of lung field (resolution of upper left lung blood vessel and bottom right lung blood vessel, and outline of the bronchial tube). Mediastinum region (Outline of trachea and the main bronchial tubes (left and right), and resolution of each margin of the mediastinum region). Granularity (soft tissue under shoulder blade). Sharpness (trabecular of collar bone and margin of rib bone). Contrast (balance of entire density). For physical and visual evaluation, we used the optimized Virtual Grid parameter to PMMA on account of correcting the different scatter characteristics from human tissue. [Materials] #Diagnostic x-ray high voltage unit: UD150B-40 (manufactured by SHIMADZU) #Ion chamber dosimeter: 2026C (manufactured by TOYO MEDIC) #FPD System: CALNEO C 1417 Wireless SQ (manufactured by FUJIFILM) and CONSOLE ADVANCE Ver.8.0 (manufactured by FUJIFILM) #Phantom: Contrast phantom (scratchbuilt), CDRAD phantom (manufactured by Artinis Medical Systems) and Page 6 of 25
Chest phantom (manufactured by KYOTO KAGAKU) #Analysing software: Image J and CDRAD Analyser Version 2.1 (manufactured by Artinis Medical Systems) Images for this section: Table 1 Page 7 of 25
Fig. 3 Page 8 of 25
Fig. 4 Page 9 of 25
Results [1] Physical evaluation The level of image contrast for increasing thickness of PMMA at 60kVp and 100% dose is shown in Fig.5. Images exposed without a grid (-), had the lowest levels of contrast when compared to the contrast of images exposed with a grid (+) and the non-grid (-) images processed by Virtual Grid processing. In addition, it was found that the Virtual Grid processed images had higher contrast levels than images exposed with a grid (+). Although the contrast levels of images exposed with a grid (+) and without a grid (-) remarkably decreased as the thickness of PMMA increased from 9.5cm to 14.5cm, the contrast levels of images processed by Virtual Grid were more stable and did not decrease to the same extent. The level of image contrast for increasing thickness of PMMA at 80kVp and 100% dose is shown in Fig.6. As with the results for 60kVp, the contrast of images exposed with a grid (+) and the non grid (-) images processed by Virtual Grid were improved compared with the images exposed without a grid (-). As well the contrast levels of Virtual Grid processing images were higher and more stable and did not decrease to the same extent over the increased PMMA thickness as the non grid (-) images and grid (+) images. The noise characteristics of images acquired at the x-ray tube voltage of 60kVp and 100% exposure dose are shown in Fig.7. The graph shows the noise value (N = (SD0#SD10) / 2) against increasing thickness of PMMA. Because the use of a grid (+) decreases the x-ray dose reaching the FPD, the granularity is greater and the noise value is higher; therefore, images exposed with a grid (+) showed significantly higher noise values than images exposed without a grid (-). The characteristics of the noise in the non-grid (-) images processed with Virtual Grid processing was superior to images exposed with a grid (+) and closer to those values of the non-grid (-) images. At the 9.5cm PMMA thickness, the characteristics of the noise in images processed by Virtual Grid processing were slightly better than images exposed without a grid (-). Fig.8 shows the noise characteristics for images acquired at 80kVp and 100% exposure dose. As with the results for 60kVp and 100% exposure dose, images exposed with a grid (+) showed higher noise values when compared to images exposed without a grid (-). The characteristics of the noise in the non-grid (-) images processed with Virtual Grid processing was superior to images exposed with a grid (+) and again closer to those values of the non-grid (-) images. Page 10 of 25
CNR results for images acquired with the x-ray tube voltage of 60kVp and 100% exposure dose are shown in Fig.9. As expected CNR decreases for all image sets as the PMMA thickness increases. However, the CNR of the non-grid (-) images processed by Virtual Grid processing showed significantly higher CNR values compared with images exposed with a grid (+) and without a grid (-). The CNR results for images acquired with the 80kVp x-ray tube voltage and 100% exposure dose are shown in Fig.10. Although the overall CNR values decreased compared with the images acquired at 60kVp and 100% dose, the images processed by Virtual Grid processing still show significantly higher CNR than images exposed with a grid (+) and without a grid (-). Fig.11 shows a comparison of the CNR for images acquired with a grid (+) at 100% dose level, and non-grid images processed with Virtual Grid processing acquired at 3 dose levels (100%, 50% and 25%). The CNR in images processed by Virtual Grid processing even at 25% of the original dose was higher than images exposed with a grid at 100% dose. [2] Visual evaluation The CDRAD IQFinv results for 9.5cm PMMA thickness, x-ray tube voltages of 60kVp and 80kVp are shown in Fig.12. The x axis shows the dose difference before and after transmission through the phantom as absorbed x-ray dose. When images acquired at the x-ray tube voltages of 60kVp and 80kVp were compared with identical x-ray absorption, images processed by Virtual Grid processing showed significantly higher IQFinv values meaning improved detail visibility than in images exposed with a grid (+) and without a grid (-). The CDRAD IQFinv results for 14.5cm PMMA thickness, x-ray tube voltages of 60kVp and 80kVp are shown in Fig.13. Although the thickness of PMMA is increased, non-grid (-) images processed by Virtual Grid processing again showed higher IQFinv values and therefore improved detail visibility than images exposed with a grid (+) and without a grid (-), as observed with the PMMA thickness of 9.5cm. As a result, the visibility of details in non-grid (-) images processed by Virtual Grid processing were regarded as having a higher image visibility than images acquired with a grid (+) and without a grid (-) while reducing the absorbed dose. [3] Chest phantom Fig.14 shows the results of the Chest Phantom images acquired at an x-ray tube voltage of 80kVp and exposure dose of 1.8mAs. Non-grid (-) images processed by Virtual Grid processing were superior to images exposed with a grid (+) in the visualisation of the structures of the peripheral side of lung field, mediastinum side of lung field, mediastinum region and sharpness. Additionally, the levels of granularity and contrast in images processed by Virtual Grid processing were equivalent to images exposed with a grid (+). Page 11 of 25
[4] Discussion From the evaluative results for CNR and IQFinv, it is shown that at 100% dose non-grid images processed by Virtual Grid processing showed a level of image contrast equivalent to or greater than images exposed with a conventional secondary radiation grid (+). Even at half the dose non-grid images processed with Virtual Grid Processing showed equivalent detail visibility to images exposed with a grid (+) at 100% dose. Images exposed with a grid (+) have improved contrast, but decreased granularity characteristics compared to images exposed without a grid (-). After having said that both images exposed with a grid (+) and without a grid (-) showed equivalent levels of CNR and IQFiv in terms of results. In contrast, non-grid (-) images processed by Virtual Grid processing showed superior CNR and IQFinv to images exposed with a grid (+) and without a grid (-) because the processing allows improvement of contrast and does not decrease granularity. From the results obtained in the study it was concluded that visualisation of lung field and mediastinum regions in non-grid (-) chest images processed by Virtual Grid processing was equivalent or greater than in images exposed with a conventional secondary radiation grid (+). Figs.15 and 16 show a chest image of a patient exposed without a grid (-) using the current examination parameters of 65kVp and exposure dose of 5mAs, and an image of the identical patient processed by Virtual Grid processing at 80kVp and the exposure dose of 2mAs. The use of Virtual Grid processing has allowed for improvement of contrast in the chest image normally associated with the use of a grid (+), but in conjunction with a reduction of incident surface dose of 33% while maintaining high quality diagnostic information. Images for this section: Page 12 of 25
Fig. 5 Page 13 of 25
Fig. 6 Page 14 of 25
Fig. 7 Page 15 of 25
Fig. 8 Page 16 of 25
Fig. 9 Page 17 of 25
Fig. 10 Page 18 of 25
Fig. 11 Page 19 of 25
Fig. 12 Page 20 of 25
Fig. 13 Page 21 of 25
Fig. 14 Page 22 of 25
Fig. 15 Page 23 of 25
Fig. 16 Page 24 of 25
Conclusion The use of an FPD and new Virtual Grid Processing software was evaluated for mobile chest images. The study determined that the use of an FPD and Virtual Grid processing allowed non-grid (-) images to be acquired with equivalent or greater diagnostic information than that obtained with the use of a conventional grid (+). In addition, non-grid (-) images acquired at a reduced dose and processed with Virtual Grid processing showed equivalent detail visibility to images exposed with a conventional grid (+) at 100% dose. In Osaka General Medical Center, the incident surface dose to the patients was reduced by 33% using the exposure condition (80kVp and 2 mas) with Virtual Grid processing compared to the conventional low tube voltage exposure condition (65kVp and 5 mas). Personal information References * Japanese Society of Radiological Technologies. "Clinical Radiographic Technology Study Handbook, Vol.1", (1996): 242-248 Page 25 of 25