Digital Chest Radiography with a Selenium-Based Flat-Panel Detector Versus a Storage Phosphor System: Comparison of Soft-Copy Images

Jin Mo Goo 1 Jung-Gi Im 1 Jong Hyo Kim 1 Joon Beom Seo 2 Tae Sung Kim 3 Sang June Shine 1 Whal Lee 1 Received February 11, 2000; accepted after revision March 23, 2000. 1 Department of Radiology, Seoul National University College of Medicine and Institute of Radiation Medicine, SNUMRC, 28, Yongon-dong, Chongro-gu, Seoul 110-744, Korea. Address correspondence to J.-G. Im. 2 Department of Radiology, Gil Medical Center, Gachon Medical School, 1198, Kuwol-dong, Namdong-gu, Inchon 405-220, Korea. 3 Department of Radiology, Samsung Medical Center, College of Medicine, Sungkyunswan University, 50, Ilwon-Dong, Kangnam-Ku, Seoul 135-710, Korea. AJR 2000;175:1013 1018 0361 803X/00/1754 1013 American Roentgen Ray Society Digital Chest Radiography with a Selenium-Based Flat-Panel Detector Versus a Storage Phosphor System: Comparison of Soft-Copy Images OBJECTIVE. We compared the soft-copy images produced by a digital chest radiography system that uses a flat-panel X-ray detector based on amorphous selenium with images produced by a storage phosphor radiography system for the visualization of anatomic regions of the chest. MATERIALS AND METHODS. Two chest radiologists and two residents analyzed 46 pairs of posteroanterior chest radiographs on high-resolution video monitors (2560 2048 8 bits). In each pair, one radiograph was obtained with a storage phosphor radiography system, and the other radiograph was obtained with a selenium-based flat-panel detector radiography system. Each pair of radiographs was obtained at the same exposure settings. The interpreter rated the visibility and radiographic quality of 11 different anatomic regions. Each pair of images was ranked on a fivepoint scale (1 = prefer image A, 3 = no preference, 5 = prefer image B) for preference of technique. Statistical significance of preference was determined using the Wilcoxon s signed rank test. RESULTS. The interpreters had a statistically significant preference for the selenium-based radiography system in six (unobscured lung, hilum, rib, minor fissure, heart border, and overall appearance) of 11 anatomic regions (p < 0.001) and for the storage phosphor system in two regions (proximal airway and thoracic spine) (p < 0.05). Chest radiologists strongly preferred seleniumbased images in eight regions, and they did not prefer storage phosphor images in any region. CONCLUSION. The soft-copy images produced by the selenium-based radiography system were perceived as equal or superior to those produced by the storage phosphor system in most but not all anatomic regions. M any studies have addressed the feasibility of using a picture archiving and communication system (PACS) as an alternative to film-based radiology [1 4]. There are many advantages to using PACS, including reduced film or processing costs and easier archiving and networking of images. Digital radiography is a suitable and important digital technique for the implementation of PACS because conventional radiographs and other projection radiographs are the most frequently obtained images in the diagnostic radiology department. The most important clinical criterion for use of the PACS is the ability to achieve acceptable accuracy when interpreting radiologic images at a soft-copy workstation. Most previous reports focused on comparing soft-copy images with hard-copy images [1 8] or comparing hard-copy images with digital and film-screen images [9 13]. During recent years, with the rapid development of electronic and computer technology, digital radiologic detectors have undergone considerable investigation and development. A new technology, a flat-panel digital detector that uses active matrix readout of amorphous selenium, has been proposed [14 18]. The active matrix consists of a two-dimensional array of thin-film transistors. In comparison with filmscreen and storage phosphor radiography systems, the selenium detector is characterized by a higher detective quantum efficiency [19]. Anticipating that the image quality of the selenium detector system exceeds that of film-screen and storage phosphor radiography systems, the new system could allow an improved detection of small, low-contrast lesions. We compared the soft-copy images of a new digital chest radiography system that uses a flat-panel X-ray detector based on amorphous selenium with those of a storage phosphor radiography system for the visualization of anatomic regions of the chest. AJR:175, October 2000 1013

Goo et al. Materials and Methods Digital Detector System Description Posteroanterior chest radiographs were obtained using two digital detector systems. The storage phosphor images were obtained with an FCR-9000 unit (Fuji, Tokyo, Japan). We used 35 43 cm ST-V imaging plates (Fuji) with a matrix of 1760 2140 10 bits and a pixel size of 0.2 mm. The selenium-based images were obtained with a DirectRay unit (Direct Radiography, Newark, DE) using a 35 43 cm solidstate detector. A complete description of the selenium-based flat-panel detector is presented elsewhere [14 18] and briefly outlined below. The solid-state selenium detector radiography system is based on thin-film transistor arrays and is constructed by adding amorphous selenium as the photoconductor material. Before exposure, an electric field is applied across the amorphous selenium layer through a bias electrode on the top surface of the selenium. As X-ray beams are absorbed in the detector, electrons and holes are released in the selenium; and because of the electric field in the selenium, electric charges are drawn directly to the charge-collecting electrodes, where the charges are stored by a capacitor. After exposure, the capacitors are read and the signals are digitized. The electric charges of the capacitors are read in a 2560 3072 matrix (139 139 µm per pixel). The current detector implementation is a seamless mosaic of two 35 21.5 cm panels. Data are digitized into 12 bits per pixel and are collected simultaneously from each panel by an array processor. The design configuration is such that the detector integrates into existing general radiographic equipment of a Bucky stand without significant modification. The detector array is connected to the array controller by a long cable. Image Acquisition and Display Forty-six consecutive patients underwent imaging with the selenium-based and the storage phosphor TABLE 1 Region Comparison of Selenium-Based Radiography with Storage Phosphor Radiography for Revealing 11 Anatomic Regions No. of Observations for Rating of 1 2 3 4 5 Unobscured lung 38 105 34 7 0 <0.001 Hilum 7 51 119 7 0 <0.001 Minor fissure 14 51 112 7 0 <0.001 Rib 2 56 119 7 0 <0.001 Heart border 5 48 117 14 0 <0.001 Retrocardiac lung 1 32 120 31 0 0.72 Subdiaphragmatic lung 0 25 136 23 0 0.78 Azygoesophageal recess 1 27 119 36 1 0.29 Proximal airway 1 22 118 43 0 <0.05 Thoracic spine 0 26 106 51 1 <0.005 Overall appearance 10 85 79 10 0 <0.001 Note. Composite data from 2024 observations by four observers. Rating of 1 = strongly preferred selenium-based image, 2 = somewhat preferred selenium-based image, 3 = no preference, 4 = somewhat preferred storage phosphor image, 5 = strongly preferred storage phosphor image. a Wilcoxon s signed rank test. p a systems using an institutional review board approved protocol; patient consent was obtained in all cases. Patients with an opacity occupying one third of their hemithorax on previous radiographs or with a history of thoracic surgery were excluded from the study. The patient group was primarily composed of inpatients, 43% of whom were women. The patients had radiographs with normal findings (n = 15) or radiographs with one or more abnormalities (n = 31). Two Bucky stands were set up on opposite sides of the same room for each radiography system. Radiography was performed using the same tube generator. The same radiographer performed both examinations at the same exposure settings: 100 kv and 8.0 mas using a 180-cm focus-detector distance. Both imaging systems included a moving 10:1 antiscatter grid (103 lines per inch). The digital data were sent to a PACS server (Radmax; MaroTech, Seoul, Korea) and distributed to workstations. All images were downloaded onto a local hard drive of a display workstation before interpretation. Each storage phosphor image was 7.18 MB, and each selenium-based image was 15.0 MB. Two 21-inch video monitors with 2048 2560 8-bit pixels (DR110; Dataray, Denver, CO) were used in a darkened room for side-by-side image display. The monitor operated at 71 Hz in an interlaced mode. The monitor had a maximum brightness level of 100 footlamberts. Because the video monitors we used have a matrix of 2048 2560 pixels, only part of the selenium-based image data (2560 3072 matrix) was shown using these displays. The selenium-based image matrices were reduced by 20% for display on the workstation. Gray-scale digital images were modified by means of a 12- to 8-bit (selenium-based image) or 10- to 8-bit (storage phosphor image) look-up table. To eliminate the difference between the two monitors, the maximum brightness of the two monitors was adjusted to be equal. The soft-copy images were displayed without unsharp masking. Only the window width and window level of the images were automatically optimized by the customized program. Interpreters were allowed to adjust the brightness and contrast of the images. Magnification of the images was not allowed. For this interpreter preference study, patient identification was obscured on all images and replaced by a sequence number. Each pair of images was displayed side-by-side in a random manner. Data Analysis Four radiologists compared the paired images using a standardized protocol. Two of the four interpreters were chest radiologists who had fellowship training in thoracic imaging and whose clinical practice involved almost exclusively chest imaging; the other two interpreters were senior residents. All interpreters were accustomed to using PACS in daily practice. The images were interpreted independently, and each interpreter was blinded to patient history. Eleven anatomic regions were evaluated in the posteroanterior views. The regions were the unobscured lung, hilum, minor fissure, retrocardiac lung, lung projected below the diaphragm (subdiaphragmatic lung), azygoesophageal recess, heart border, rib, proximal airway, thoracic spine, and overall appearance (Figs. 1 and 2). Each pair of images was ranked on a scale from one to five (1 = strongly preferred A, 2 = somewhat preferred A, 3 = no preference, 4 = somewhat preferred B, and 5 = strongly preferred B [technique A = selenium-based radiography system]) for preference of technique. These responses were recorded and resorted to each system for statistical analysis. The difference caused by minor position changes (e.g., minor fissure) was ignored, and regions were analyzed as they appeared (Fig. 3). The interpreters responses were evaluated in three groups: responses of chest radiologists, responses of residents, and responses of all interpreters. To determine interpreter preference for each imaging finding, statistical significance was calculated using the Wilcoxon s signed rank test at the conventional level (p < 0.05) for each of the 11 regions. Results The composite data for all interpreters for each anatomic region are summarized in Table 1. The interpreters had a statistically significant preference for the selenium-based system in six (unobscured lung, hilum, minor fissure, rib, heart border, and overall appearance) of 11 anatomic regions (p < 0.001) and for the storage phosphor system in two regions (proximal airway and thoracic spine) (p < 0.05). Chest radiologists strongly preferred selenium-based images in eight regions (unobscured lung, hilum, minor fissure, rib, heart border, retrocardiac lung, subdiaphragmatic lung, and overall appearance) (p < 0.005); they did not prefer the storage phosphor images in any region. Residents preferred selenium-based images in five regions (unobscured lung, hilum, minor fissure, rib, and over- 1014 AJR:175, October 2000

Digital Chest Radiography with Selenium or Phosphor all appearance) (p < 0.001); they preferred the storage phosphor images in two regions (proximal airway and thoracic spine) (p < 0.05). Discussion A number of articles describe the current experience and future potential of digital imaging network and display stations; they address the use of PACS for image interpretation and for clinical consultation between radiologists and clinicians [1, 2, 4, 5, 20]. Projection radiography has been the last roadblock to achieving a totally digital radiology department. At the present time, storage phosphor digital imaging may be the most appropriate Fig. 1. 52-year-old man with colon cancer and multiple pulmonary metastatic nodules. A and B, Selenium-based (A) and storage phosphor (B) images show unobscured retrocardiac and subdiaphragmatic lung. Note multiple nodules and linear opacities. A B technique for entry of images into a PACS in which images stored in image management systems can be displayed on high-resolution monitors [8, 21]. Recently, a new class of digital chest imaging systems based on selenium detectors has been introduced [7 9, 15, 22 24]. Several studies were performed to compare selenium-based radiography with film-screen radiography or storage phosphor radiography, and selenium-based radiography was reported to be superior or equivalent to filmscreen or storage phosphor systems for the visualization of normal anatomy or abnormalities [7 9, 22 24]. However, these studies were performed with drum-type selenium radiography. Our results revealed that interpreters preferred selenium-based images in six of 11 anatomic regions. These regions are confined to nonmediastinal areas. Although the interpreters were not told which images were from the selenium-based system and which were from the storage phosphor system, the features of nonmediastinal areas, especially the unobscured lung, were so conspicuous in selenium-based images that the interpreters could distinguish between the two systems. This factor is a potential source of bias in our study. For example, bias could have influenced our results if the interpreter had a prejudiced opinion and preferred one radiography system over the other. One of the possible causes for the difference in image quality is the difference in spatial resolution. Pixel size is an important parameter in digital radiography because it directly influences both the spatial resolution of the images and the cost of the imaging system. Other factors being equal, the smaller the pixel size, the better the image quality, particularly in the depiction of fine detail. Currently, most radiologists agree that a spatial sampling resolution of 5 pixels/mm (0.2-mm pixels) in an adult chest image is required for the depiction of fine detail such as septal lines or a subtle pneumothorax [13, 25, 26]. In our study, the pixel size of the storage phosphor system was 0.2 mm, and the pixel size of the selenium-based system was 0.139 mm. Theoretically, the spatial resolution of the selenium-based system was 1.4 times better than that of storage phosphor system. However, considering the limitation of the monitor in displaying images, the factor decreases to 1.2. Within certain limits, a smaller pixel size allows improved spatial resolution in a digital image. Conversely, use of a larger pixel size (i.e., fewer data) facilitates storage, AJR:175, October 2000 1015

Goo et al. Fig. 2. 28-year-old man with Hodgkin s disease. A and B, Selenium-based (A) and storage phosphor (B) images show mediastinal area including proximal airway and thoracic spine. A B transmission, and processing of digital information. The optimal pixel size for a given application will be one that is just small enough to provide an acceptable level of diagnostic accuracy. Then adequate image quality will be maintained while data storage requirements are reduced to a practical minimum. Although currently a new generation of storage phosphor devices with higher resolution 4K image formats is available, the data size of the 4K system is too large to use in the PACS available at our hospital. This is an important factor limiting the clinical implementation of a PACS. The data size of selenium-based images (15.0 MB) was more than twice the size of the storage phosphor images (7.18 MB). However, recent research [27] suggests that data compression techniques that reduce the amount of digital data by as much as 25-fold can be used with a minimum loss of perceptible quality. More pixels do not always mean higher spatial resolution because image blurring can result from the scatter of X-ray beams, scatter of light, or both in the detector. In a selenium-based flatpanel system, the light or electric charges are converted into digital data by arrays of semiconductor elements. This absorbs the X-ray photons and directly produces electric charges that are attracted to the plate electrodes by an applied potential difference. Therefore, the conversion of X-ray photons to electric charges is direct and does not use an intervening light stage as in an intensifying screen system (filmscreen radiography) or photostimulable phosphor (storage phosphor radiography system. The light is scattered in the intensifying screen or photostimulable phosphor producing a curved signal profile that blurs the image. The selenium-based system, without this intermediate light fluorescence, has a square profile signal, providing sharper image quality [28]. The number of gray levels in a digital system determines how well it reproduces subtle variations in attenuation in an accurate and visually acceptable way. Unlike analog images, a digital image has only a predetermined number of shades of gray. In practice, 256 shades of gray are the minimum number necessary for adequately displaying a digital chest radiograph; 1024 shades (10 bits) or, preferably, 4096 shades (12 bits) provide a visibly superior display [29]. In our study, interpreters preferred storage phosphor images in two regions (proximal airway and thoracic spine) that were located in the mediastinal area. Because the spatial resolution is superior in selenium-based images, this difference may be caused by the contrast scaling of an image. The contrast scaling refers to a series of lookup tables that can be applied to the digital image data to map pixel values to luminance of monitor display. The selenium-based image (12 bits) requires more contrast scaling than the storage phosphor image (10 bits), and the default setting on a monitor may not be adequate for displaying the mediastinum in a selenium-based image. Although this could be overcome by adjusting the contrast on a workstation, interpreters are not likely to adjust the contrast when comparing images. Another explanation for the difference is that these two 1016 AJR:175, October 2000

Digital Chest Radiography with Selenium or Phosphor regions are characterized by medium-sized low-contrast shadows in a highly attenuated area that does not require small pixels to preserve sharp edges. These areas have a relatively low signal-to-noise ratio because of the quantum mottles produced by the highly attenuated X-ray beam. When comparing the images of this area produced by the seleniumbased system with those produced by the storage phosphor system, the quantum mottles were more notable on the selenium-based images. On the storage phosphor images, the quantum mottles might be blurred in the light scattering process, which does not occur in the selenium-based system. Therefore, the light scattering process in the storage phosphor system might have worked as an image filter that increased the apparent signal-tonoise ratio of the shadows in the two regions. This study was a first step in the evaluation of the diagnostic use of selenium-based flatpanel chest radiography because interpreter preference is an important factor in its clinical acceptance. In general, chest radiologists preferred selenium-based images more than residents did. Chest radiologists had no statistically significant preference for storage phosphor images. Both groups strongly preferred seleniumbased images in overall appearance. However, further receiver operating characteristic studies are necessary to compare the diagnostic accuracy of selenium-based images with that of storage phosphor images. Further work is needed to optimize anatomy-specific exposure techniques. In our study, voltage and amperage were identical for the selenium-based and storage phosphor radiography systems. In our study, we used 100 kvp because this is the voltage used at our hospital. Further improvement in the quality of digital radiographic images and a possible reduction in X-ray dose may be possible by optimizing the voltage and amperage of flat-panel detectors. When we used a stationary grid, corduroy artifacts developed because of image aliasing; therefore, we used a moving grid to avoid these artifacts. Although the motor required for a grid may interfere with the electronics, there were no significant artifacts in our series. In conclusion, the soft-copy images of a selenium-based flat-panel radiography system are perceived as equal to or superior to those of a storage phosphor radiography system in most but not all anatomic regions. This new digital radiography system could be an appropriate technique for the entry of images into a PACS. Fig. 3. 42-year-old man with lung cancer and malignant pleural effusion. A and B, Selenium-based (A) and storage phosphor (B) images show unobscured lung, hilum, and ribs. Sharp right minor fissure is seen on selenium-based image; fissure is obscured on storage phosphor image. Although this difference may be caused by minor position difference of patients, images were evaluated as they appeared. References 1. Razavi M, Sayre JW, Taira RK, et al. Receiveroperating-characteristic study of chest radiographs in children: digital hard-copy film vs 2K 2K soft-copy images. AJR 1992;158:443 448 2. Ishigaki T, Endo T, Ikeda M, et al. Subtle pulmonary disease: detection with computed radiography versus conventional chest radiography. Radiology 1996;201:51 60 3. Hayrapetian A, Aberle DR, Huang JK, et al. Comparison of 2048-line digital display formats in conventional radiographs: an ROC study. AJR 1989;152:1113 1118 4. Slasky BS, Gur D, Good WF, et al. Receiver operator characteristic analysis of chest image inter- A B AJR:175, October 2000 1017

Goo et al. pretation with conventional, laser-printed, and high-resolution workstation images. Radiology 1990;174:775 780 5. Kundel HL, Gefter W, Aronchick J, et al. Accuracy of bedside chest hard-copy screen-film versus hardand soft-copy computed radiographs in a medical intensive care unit: receiver operating characteristic analysis. Radiology 1997;205:859 863 6. Thaete FL, Fuhrman CR, Loiver JH, et al. Digital radiography and conventional imaging of the chest: a comparison of observer performance. AJR 1994;162:575 581 7. Cox GG, Cook LT, McMillan JH, Rosenthal SJ, Dwyer SJ III. Chest radiography: comparison of high-resolution digital displays with conventional and digital film. Radiology 1990;176:771 776 8. Frank MS, Jost RG, Molina PL, et al. High-resolution computer display of portable, digital chest radiographs of adults: suitability for primary interpretation. AJR 1993;160:473 477 9. Kido S, Ikezoe J, Takeuchi N, et al. Interpretation of subtle interstitial lung abnormalities: conventional versus storage phosphor radiography. Radiology 1993;187:527 533 10. van Heesewijk HPM, Neitzel U, van der Graaf Y, de Valois JC, Feldberg MAM. Digital chest imaging with a selenium detector: comparison with conventional radiography for visualization of specific anatomic regions of the chest. AJR 1995;165:535 540 11. Floyd CE Jr, Baker JA, Chotas HG, Delong DM, Ravin CE. Selenium-based digital radiography of the chest: radiologists preference compared with film-screen radiographs. AJR 1995;165:1353 1358 12. Schaefer-Prokop CM, Prokop M, Schmidt A, Neitzel U, Galanski M. Selenium radiography versus storage phosphor and conventional radiography in the detection of simulated chest lesions. Radiology 1996;201:45 50 13. MacMahon H, Vyborny DJ, Metz CE, et al. Digital radiography of subtle pulmonary abnormalities: an ROC study of the effect of pixel size on observer performance. Radiology 1986;158:21 26 14. Zhao W, Rowlands JA. X-ray imaging using amorphous selenium: feasibility of a flat panel self-scanned detector for digital radiology. Med Phys 1995;22:1595 1604 15. Zhao W, Rowlands JA. Digital radiology using active matrix readout of amorphous selenium: theoretical analysis of detective quantum efficiency. Med Phys 1997;24:1819 1833 16. Zhao W, Blevis I, Germann S, Rowlands JA, Waechter D, Huang Z. Digital radiology using active matrix readout of amorphous selenium: construction and evaluation of a prototype real-time detector. Med Phys 1997;24:1834 1843 17. Chotas HG, Dobbins JT, Ravin CE. Principles of digital radiography with large-area, electronically readable detectors: a review of the basics. Radiology 1999;210:595 599 18. Piraino DW, Davros WJ, Lieber M, et al. Selenium-based digital radiography versus conventional film-screen radiography of the hands and feet: a subjective comparison. AJR 1999;172: 177 184 19. Neitzel U, Maack I, Gunther-Kohfahl S. Image quality of a digital chest radiography system based on a selenium detector. Med Phys 1994; 21:509 516 20. Kundel HL, Seshadri SB, Langlotz CP, et al. Prospective study of a PACS: information flow and clinical action in a medical intensive care unit. Radiology 1996;199:143 149 21. Glazer HS, Muka E, Sagel SS, Jost RG. New techniques in chest radiography. Radiol Clin North Am 1994;32:711 729 22. van Heesewijk HPM, van der Graaf Y, de Valois JC, Vos JA, Feldberg MAM. Chest imaging with a selenium detector versus conventional film radiography: a CT-controlled study. Radiology 1996; 200:687 690 23. Woodard PK, Slone RM, Gierada DS, Reiker GG, Pilgram TK, Jost RG. Chest radiography: depiction of normal anatomy and pathologic structures with selenium-based digital radiography versus conventional screen-film radiography. Radiology 1997;203:197 201 24. Woodard PK, Slone RM, Sagel SS, et al. Detection of CT-proved pulmonary nodules: comparison of selenium-based digital and conventional screen-film chest radiographs. Radiology 1998; 209:705 709 25. Ravin CE, Harrell G, Chotas MS. Chest radiography. Radiology 1997;204:593 600 26. Seeley GW, Fisher HD, Stempski MO, et al. Total digital radiology department: spatial resolution requirements. AJR 1987;148:421 426 27. MacMahon H, Doi K, Sanada S, et al. Effect of data compression on diagnostic accuracy in digital chest radiography: receiver-operating characteristic study. Radiology 1991;178:175 179 28. Rowlands JA, Zhao W, Blevis IM, Waechter DF, Huang Z. Flat-panel digital radiology with amorphous selenium and active-matrix readout. Radio- Graphics 1997;17:753 760 29. MacMahon H, Vyborny C. Technical advances in chest radiography. AJR 1994;163:1049 1059 1018 AJR:175, October 2000