757 ノート CT 1 奥村英一郎神前 6 久保美華代 2 僚 7 白坂菜摘 3 宮下慶太 8 武田大樹 4 上田梨菜 9 橋本憲幸 5 上舞優介 2013 10 10 2014 6 5 Code Nos. 520 532, 533 1 2 3 4 5 6 7 8 9 EIZO Diagnostic Detection Performance of a Simulated Nodule in Chest Computed Tomography Images and Gray and Color Nuclear Medicine Images: Comparison between a Medical Liquid Crystal Display Monitor and an Ordinary Liquid Crystal Display Monitor Eiichiro Okumura, 1 Riyou Kamimae, 2 Kenta Miyashita, 3 Rina Ueda, 4 Yusuke Kanmae, 5 Mikayo Kubo, 6 Natsumi Shirasaka, 7 Taiki Takeda, 8 and Noriyuki Hashimoto 9 1 Department of Medical Radiological Technology, Kagoshima Medical Technology College 2 Department of Radiology, Nagasaki Kawatana Medical Center 3 Department of Radiology, Osaki Midtown Medicare Clinic 4 Department of Radiology, Himeji Medical Center 5 Center for Diagnostic Imaging, Imamura Hospital 6 Department of Radiology, Hujimoto General Hospital 7 Department of Radiology, Ureshino Medical Center 8 Department of Radiology, Imusu Yokohama Kariba Neurosurgical Hospital 9 EIZO Corporation Received October 10, 2013; Revision accepted June 5, 2014 Code Nos. 520, 532, 533 Summary The purpose of this study was to evaluate the detection performance of simulated nodules in chest computed tomography (CT) images and nuclear medicine images with an ordinary liquid crystal display (LCD) and a medical LCD (grayscale standard display function: GSDF) and gamma 2.2. We collected 72 chest CT image slices obtained from an LSCT phantom with simulated signals composed of various sizes and CT values and 78 slices of monochrome and color nuclear medicine images obtained from a digital phantom with a simulated signal composed of various sizes and radiation levels. Six observers performed receiver operating characteristic (ROC) analysis using a continuous scale. The area under the ROC curve (AUC) was calculated for each monitor. The average AUC values for detection of chest CT images on a medical LCD (GSDF), medical LCD (gamma 2.2), and ordinary LCD were 0.71, 0.67, and 0.73, respectively. The average AUC values for detection of monochrome nuclear medicine images using a medical LCD (GSDF), medical LCD (gamma 2.2), and ordinary LCD were 0.81, 0.75, and 0.72, respectively. The average AUC values for detection of color nuclear medicine images on a medical LCD (GSDF), medical LCD (gamma 2.2), and ordinary LCD were 0.88, 0.86, and 0.90, respectively. Observer performance for detection of simulated nodules in chest CT images and nuclear medicine images was not significantly different between the three LCD monitors. We therefore conclude that an ordinary LCD monitor can be used to detect simulated nodules in chest CT images and nuclear medicine images. Key words: liquid crystal display (LCD), receiver operating characteristic (ROC) analysis, chest computed tomography (CT) image, nuclear medicine images, simulated nodule *Proceeding author
758 緒言 / / liquid crystal display LCD 1 X 2 6 7 9 X grayscale standard display function GSDF 3 mega M pixels LCD GSDF 5 M pixels LCD 10 JESRA JESRA X-0093 * A -2010 1 LCD 11 LCD X computed tomography CT magnetic resonance imaging MRI 512 512 64 64 1 M pixels 2 M pixels 10 GSDF LCD 2.2 LCD LCD LCD 12, 13 14 CT Kawasumi 14 CT LCD GSDF LCD 2.2 CT CT CT 1 M pixels LCD 2.2 2 M pixels LCD GSDF 2.2 3 receiver operating characteristic ROC 3 CT 3 1. 方法 1-1 Table 1 LCD 2 M pixels EIZO RadiForce RX220 LCD 1 M pixels EIZO FlexScan L797 2 LCD LCD GSDF 2.2 LCD 0.7 cd/m 2 400 cd/m 2 LCD 2.2 RadiCS X1 EIZO LS-100 LCD LCD Fig. 1 2 CS1000A 18 Table 2 red-green-blue RGB 39 Fig. 3 5 15 17
759 Table 1 Display parameters for medical LCD and ordinary LCD monitors on the observer test 2 M pixels medical LCD monitor 1 M pixels general LCD monitor (RadiForce RX220) (FlexScan L797) Panel size 54 cm 48 cm Matrix size of monitor 1200 1600 1280 1024 Dot pitch 0.270 mm 0.294 mm Display bit-depth 10 bits 10 bits color Luminance setting range 0.7 400 cd/m 2 Maximum luminance 376 cd/m 2 403.2 cd/m 2 284.5 cd/m 2 Minimum luminance 0.42 cd/m 2 0.71 cd/m 2 0.61 cd/m 2 Display function GSDF γ 2.2 γ 2.2 Fig. 1 GSDF and ganma characteristic curve for medical and ordinary LCD monitors. Fig. 2 ΔL/L for a JND of medical and ordinary LCD monitors. Table 2 Color code As shown Fig. 3, we painted the central point of color scale test pattern with the color combination of these color codes. Gray No.1 No.8 No.1 No.2 No.3 No.4 No.5 No.6 No.7 No.8 Red 32 64 96 128 160 192 224 255 Green 32 64 96 128 160 192 224 255 Blue 32 64 96 128 160 192 224 255 Red No.1 No.4 Green No.1 No.4 No.1 No.2 No.3 No.4 No.1 No.2 No.3 No.4 Red 128 192 255 255 0 64 0 128 Green 0 64 0 128 128 192 255 255 Blue 0 64 0 128 0 64 0 128 Blue No.1 No.4 Yellow No.1 No.4 No.1 No.2 No.3 No.4 No.1 No.2 No.3 No.4 Red 0 64 0 128 128 128 255 255 Green 0 64 0 128 128 128 255 255 Blue 128 192 255 255 255 0 0 128 Magenta No.1 No.4 Cyan No.1 No.4 No.1 No.2 No.3 No.4 No.1 No.2 No.3 No.4 Red 128 192 255 255 0 64 0 128 Green 0 64 0 128 128 192 255 255 Blue 128 192 255 255 128 192 255 255
760 Fig. 3 Color scale test pattern. 1-2 1-2-1 CT CT Subrina CYON LSCT LSCT0001 CT 120 kv 50 ma second mas 10 mm pitch 2.0 10 mm 18 20 CT 10 mas 200 mas pitch 1.0 2.5 CT 4 12 mm 5 2 10 mm 5 100 hounsfield unit HU 270 HU Fig. 6 a f 120 kv 10 mm pitch 1.0 10 mas 200 mas Fig. 6 g i 120 kv 10 mm 20 mas pitch 1.5 2.5 Fig. 6 a f Fig. 6 g i pitch Fig. 6 Fig. 7 noise Fig. 4 Fig. 5 The CIE u v space on a grayscale test pattern for (a) medical LCD (gamma 2.2), (b) medical LCD (GSDF), and (c) ordinary LCD (gamma 2.2). The CIE u v color space on a color test pattern for medical LCD (gamma 2.2), medical LCD (GSDF), and ordinary LCD (gamma 2.2). power spectrum NPS 10 ma/s 200 ma/s Fig. 8 CT 27
761 Fig. 6 The effect of dose and pitch factor on a water phantom. Scan conditions: 120 kv, 10 mm slice thickness, pitch: 1.0, (a) 10 mas, (b) 20 mas, (c) 50 mas, (d) 100 mas, (e) 150 mas, (f) 200 mas and 120 kv, 20 mas, (g) pitch: 1.5, (h) pitch: 2.0, (i) pitch: 2.5. Fig. 8 Chest CT images of lung apex, bifurcation of the trachea, and lung base. Scan conditions: 120 kv, 10 mm slice thickness, pitch factor 1.0, (a c): 200 mas, (d f): 10 mas. Fig. 9 Structure and size of digital phantom. Fig. 7 The effect of dose on WS value. 48 CT 1-2-2 SPECT 17 18 Fig. 9 200 mm 200 mm 30 mm 60 40 20 10 4 mm 5 1 0 2 4 21 electron gamma shower 4 EGS4 single photon emission computed tomography SPECT 21 low energy high resolution LEHR low energy general purpose LEGP 200 mm 150 mm
762 Table 3 Projection data with a digital phantom in a nuclear medicine image Collimator Radius of rotation Pixel size Step angle Maximum count No.1 LEHR 200 mm 4 mm 3 100 No.2 LEGP 200 mm 4 mm 3 100 No.3 LEHR 150 mm 4 mm 3 100 No.4 LEHR 150 mm 2 mm 3 100 No.5 LEHR 200 mm 6 mm 3 100 No.6 LEGP 200 mm 6 mm 3 100 No.7 LEHR 200 mm 4 mm 3 50 2 mm 4 mm 6 mm 3 50 100 Table 3 21 RI Daemon Research Image processor Table 3 SPECT SPECT Butterworth 8 0.5 cycles/cm ramp filtered back projection FBP iteration 10 subset 2 ordered subsets expectation maximization OS-EM Fig. 10 OS-EM SPECT Fig. 10 1 SPECT 2 5 20 7 SPECT Fig. 10 1 5 60 40 20 10 4 mm Table 3 SPECT SPECT 48 30 48 5 1 5 1 2 3 4 5 1 4 12 48 0 100 rainbow gray 2 Fig. 10 SPECT reconstruction images with OS-EM method. (a) Ideal SPECT reconstruction images, SPECT reconstruction images with photoelectric peak window of (b) total, (c) true, (d) scatter, and SPECT reconstruction images with low energy of (e) sub window on regions including rods (60 4 mm). 1-3 CT 8 bits 8 bits MLF-471 30 lx a b c d e
763 Table 4 AUC values for detection of a simuled nodule in chest CT images for six observers Observer Medical LCD (γ 2.2) Medical LCD (GSDF) General LCD (γ 2.2) 1 0.694 0.651 0.778 2 0.848 0.927 0.848 3 0.563 0.522 0.546 4 0.664 0.778 0.748 5 0.691 0.655 0.651 6 0.558 0.733 0.788 Average 0.670 0.711 0.727 6 22 ROCViewerForMethod- 1Ver1.0.1 22 512 512 0.596 mm CT LCD 138.24 0.27 512 mm 0.452 138.24/305.15 LCD 150.53 0.294 512 mm 0.493 150.53/305.15 256 256 1 mm LCD 69.12 256 0.27 mm 0.27 69.12/256 LCD 75.26 256 0.294 mm 0.294 75.26/256 LCD 2 3 1-4 ROC 6 ROC area under ROC curve: AUC Metz ROCKIT 22 AUC Fig. 11 Average ROC curve for detection of simulated nodule in chest CT images. Student-t 5 ROC Metz PlotROC 22 DBM multi-reader multi-case MRMC Jackknife ROC 23, 24 2. 結果 CT LCD LCD 6 AUC Table 4 AUC LCD 2.2 0.73±0.10 LCD GSDF 0.71±0.13 LCD 2.2 0.67±0.10 6 ROC Fig. 11 LCD 2.2
764 Table 5 P-values (α=0.05) in statistical testing for comparison of displays Chest CT image Monochrome nuclear image Color nuclear image Medical LCD (γ 2.2) vs Medical LCD (GSDF) 0.332 0.116 0.506 General LCD (γ 2.2) vs Medical LCD (γ 2.2) 0.220 0.534 0.175 Medical LCD (GSDF) vs General LCD (γ 2.2) 0.612 0.001 0.440 Table 6 (a) AUC values for detection of simulated signals in (a) monochrome and (b) color nuclear medicine images for six observers Observer Medical LCD (γ 2.2) Medical LCD (GSDF) General LCD (γ 2.2) 1 0.745 0.786 0.725 2 0.660 0.820 0.768 3 0.777 0.745 0.630 4 0.818 0.835 0.719 5 0.663 0.801 0.708 6 0.816 0.841 0.777 Average 0.746 0.805 0.721 (b) Observer Medical LCD (γ 2.2) Medical LCD (GSDF) General LCD (γ 2.2) 1 0.901 0.953 0.941 2 0.797 0.815 0.924 3 0.789 0.888 0.876 4 0.909 0.907 0.894 5 0.902 0.808 0.878 6 0.885 0.924 0.898 Average 0.864 0.883 0.902 AUC Table 5 6 AUC Student-t 3 DBM MRMC 3 P=0.442 LCD LCD 6 AUC Table 6 a AUC LCD 2.2 0.72±0.05 LCD GSDF 0.81±0.03 LCD 2.2 0.75±0.06 6 ROC Fig. 12 a LCD GSDF AUC Table 5 6 AUC Student-t LCD GSDF LCD 2.2 P=0.001 2 DBM MRMC 3 P=0.097 LCD LCD 6 AUC Table 6 b AUC LCD 2.2 0.90±0.02 LCD GSDF 0.88±0.05 LCD 2.2 0.86±0.05 6 ROC Fig. 12 b LCD 2.2 AUC Table 5 6 AUC Student-t 3 DBM MRMC
765 a Fig. 12 b Average ROC curve for detection of simulated signal in (a) monochrome and (b) color nuclear medicine images. 3 P=0.425 3. 考察 LCD LCD CT LCD LCD LCD LCD X CT CT LCD LCD 400 cd/m 2 LCD 300 cd/m 2 LCD GSDF 2.2 LCD 2.2 CT GSDF X 2.2 GSDF GSDF 2.2 25, 26 LCD LCD CT 2.2 picture archiving and communication system PACS LCD GSDF CT LCD 2.2 GSDF LCD LCD 2.2 LCD 2.2 LCD LCD LCD LCD
766 Fig. 4 a b LCD No.1 0 No.18 255 LCD No.1 No.2 15 No.18 LCD LCD No.1 No.2 LCD LCD 27 LCD LCD CT Fig. 5 LCD GSDF LCD 2.2 LCD 2.2 LCD LCD 255 100 0 0 0 100 100 0 2 100 110 10 20 100 0 28 gray rainbow 28 LCD 4. 結語 1 M pixels LCD 2.2 2 M pixels LCD 2.2 GSDF CT 3 CT 謝辞 EIZO LCD 17 18 22 SPECT ROC ROCViewerForMethodVer1.0.1 7 8
767 参考文献 1 1 2002; 58(10): 1297-1302. 2 2005; 61(8): 1151-1157. 3 Langer S, Fetterly K, Mandrekar J, et al. ROC study of four LCD displays under typical medical center lighting conditions. J Digit Imaging 2006; 19(1): 30-40. 4 LCD CRT 2006; 62(5): 734-741. 5 Fukushima H, Ikeda M, Ishigaki T, et al. Influence of liquid crystal display monitors on observer performance for detection of diffuse pulmonary disease on chest radiographs. Radiat Med 2007; 25(5): 211-217. 6 Park CM, Lee HJ, Goo JM, et al. Comparison of observer performance on soft-copy reading of digital chest radiographs: high resolution liquid-crystal display monitors versus cathode-ray tube monitors. Eur J Radiol 2008; 66(1): 13-18. 7 Uematsu T, Kasami M, Uchida Y. Soft-copy reading in digital mammography of microcalcifications: diagnostic performance of a 5-megapixel cathode ray tube monitor versus a 3-megapixel liquid crystal display monitor in a clinical setting. Acta Radiol 2007; 48(7): 714-720. 8 Yamada T, Suzuki A, Uchiyama N, et al. Diagnostic performance of detecting breast cancer on computed radiographic (CR) mammograms: comparison of hard copy film, 3-megapixel liquid-crystal-display (LCD) monitor and 5-megapixel LCD monitor. Eur Radiol 2008; 18(11): 2363-2369. 9 Kamitani T, Yabuuchi H, Matsuo Y, et al. Diagnostic performance in differentiation of breast lesion on digital mammograms: comparison among hard-copy film, 3-megapixel LCD monitor, and 5-megapixel LCD monitor. Clin Imaging 2011; 35(5): 341-345. 10 2.0 2006. 11 JESRA X-0093 * A -2010 2010 6 11 12 Yoshimura K, Shimamoto K, Ikeda M, et al. A comparative contrast perception phantom image of brain CT study between high-grade and low-grade liquid crystal displays (LCDs) in electronic medical charts. Phys Med 2011; 27(2): 109-116. 13 Yamaguchi M, Fujita H, Asai Y, et al. Psychophysical analysis of monitor display functions affecting observer diagnostic performance of CT image on liquid crystal display monitors. Eur Radiol 2005; 15(12): 2487-2496. 14 Kawasumi Y, Yamada T, Ota H, et al. High-resolution monochrome liquid crystal display versus efficient household colour liquid crystal display: comparison of their diagnostic performance with unenhanced CT images in focal liver lesions. Eur Radiol 2008; 18(10): 2148-2154. 15 International Organization for Standardization (ISO), Colorimetry Part 1: Commission Imternationale de l Eclairage (IEC) standard colorimetric observers ISO Report No.11664-1:2007 (CIE, Wien, 2007). 16 International Organization for Standardization (ISO). Colorimetry Part 3: Commission Imternationale de l Eclairage (IEC) tristimulus values ISO Report No.11664-3:2012 (CIE, Wien, 2012). 17 International Organization for Standardization (ISO). Colorimetry Part 5: Commission Imternationale de l Eclairage (IEC) 1976 L*u*v* Colour space and u, v uniform chromaticity scale diagram ISO Report No.11664-5:2009 (CIE, Wien, 2009). 18 MDCT Multidetector-row CT 2007; 63(7): 809-819. 19 CT 2001; 57(9): 1089-1097. 20 5 1999. 21 2007; 63(11): 1313-1319. 22 Metz CE, Herman BA, Shen JH. Maximum likelihood estimation of receiver operating characteristic (ROC) curves from continuously-distributed data. Stat Med 1998; 17(9): 1033-1053. 23 ROC 1994; 50(10): 1726-1734. 24 ROC Jackknife 1997; 53(6): 691-698. 25 2010; 66(1): 25-32. 26 QAQC 2011; 67(3): 235-243. 27 Akamine H, Morishita J, Matsuyama M, et al. Chromaticity and correlated color temperature of the white point in medical liquid-crystal display. Med Phys 2012; 39(8): 5127-5135. 28 2008; 28(1): 13-66. 891-0133 5417-1