NEMA MITA WP 1-2017 Computed Tomography Image Quality (CTIQ) Low-Contrast Detectability (LCD) Assessment When Using Dose Reduction Technology.pdf

1、NEMA Standards PublicationNational Electrical Manufacturers AssociationNEMA/MITA WP 1-2017Computed Tomography Image Quality (CTIQ):Low-Contrast Detectability (LCD) Assessment When Using Dose Reduction TechnologyNEMA/MITA WP 1-2017 Page i 2017 National Electrical Manufacturers Association A NEMA Medi

2、cal Imaging b. No ROI will come to within 5 mm of any other rod; and c. No ROI will come to within 10 mm of the edge of the phantom. Figure 4 shows ROI size for LU and LK and ROI positions for each rod with respect to other ROIs. Figure 4 Illustrates ROI Sizes and How Far They Can Move for Each Rod.

3、 During initial testing, it was seen that the choice of rod sizes was acceptable for body protocols, but had too much contrast for useful testing with head protocols with ROC testing. That is, the AUC seen with typical head exams and no body ring was close to 0.99, which does not allow useful compar

4、isons. Two alternatives have been considered. The first is to use LROC testing with head protocols. By adding a localization parameter to the design, results allowed comparative testing. The second alternative was to create a second head phantom, with lower contrast rods. After working with the phan

5、tom manufacturer, a second phantom (head phantom) was created for testing head protocols, with the following sizes of rods, at the same positions as for the body phantom. Manufacturers were able to achieve good results with the second phantom (head phantom), but depending on the clinical protocol/po

6、pulation alternative head phantoms, such as a 16 cm phantom with a Teflon annulus to simulate bone, could be considered. NEMA/MITA WP 1-2017 Page 8 2017 National Electrical Manufacturers Association Table 1 Rod Diameters and Contrasts for Body and Head Phantoms Rod Body Phantom Head Phantom Diameter

7、 (mm) Contrast (HU) Diameter (mm) Contrast (HU) 1 3 14 2 9 2 5 7 3 5 3 7 5 4 4 4 10 3 5 3 A second concern raised during the initial testing of the phantom was quality control. Some vendors saw very different contrast (i.e., more than HU) in the rods than specified by the supplier. To help evaluate

8、the contrast, four rods were added to each phantom, with length 10 mm, but diameter 15 mm, for easier measurement of CT number. However, these rods showed a variation in CT number compared to the small rods. Therefore, follow-up work is being conducted with the phantom manufacturer with respect to t

9、he 15 mm rods. The initial version of the phantom had a 20 mm blank section, which made it difficult to get a sufficient number of blank images from a single helical scan. The second version had a 40 mm blank section, allowing more blank images to compare with a rod image. A limitation on the use of

10、 this phantom is that the contrast difference is created almost entirely by differences in density of the rod materials, rather than different rod compositions. This means that the contrast difference is relatively insensitive to changes in tube voltage. Clinically, lowering tube voltage allows impr

11、oved low-contrast detectability, since the tissues inside the body are comprised of different chemical compositions. NEMA/MITA WP 1-2017 Page 9 2017 National Electrical Manufacturers Association 6 Sample ROC Curves The following figures represent sample ROC curves with the head and body phantoms, wh

12、ich evaluate performance of images generated on actual systems. Figure 5 ROC Curve Examples. Left: 3 Mm 14 Hu Object without Body Ring. Right: 5 mm 7 Hu Object with Body Ring (Courtesy of GE Healthcare) Figure 6 LROC Curve Examples. Left: 10 Mm 3 Hu Object without Body Ring Right: 7 Mm 5 Hu Object w

13、ith Body Ring (Courtesy of GE Healthcare) NEMA/MITA WP 1-2017 Page 10 2017 National Electrical Manufacturers Association Figure 7 ROC Curve for Head Phantom (courtesy of Toshiba) 7 Conclusion The use of task-based observers to quantify low-contrast detectability as a function of dose reduction provi

14、des a practical, repeatable method for measuring performance of both FBP and IR. The phantom designed by MITA, in consultation with the FDA, is an effective tool for conducting observer studies that yields meaningful data in a manner precise enough for regulatory purposes. NEMA/MITA WP 1-2017 Page 1

15、1 2017 National Electrical Manufacturers Association Annex A Acronyms Term Acronym American College of Radiology ACR Area Under the ROC Curve AUC Channelized Hoteling Observer CHO Computed Tomography CT Computed Tomography Image Quality CTIQ Contrast to Noise Ratio CNR Figure of Merit FOM Filtered B

16、ack Projection FBP Food and Drug Administration FDA Free Response ROC FROC Iterative Reconstruction IR Localization ROC LROC Location Unknown LU Low-Contrast Detectability LCD Medical Imaging 28(2):94-108 Chakraborty, D. P. “New Developments in Observer Performance Methodology in Medical Imaging,“ S

17、eminars in Nuclear Medicine, 2011 Nov; 41(6): 401418. Hernandez-Giron 2011 med physics: 19I. Hernandez-Girn, J. Geleijns, A. Calzado, and. J. H. Veldkamp, “Automated assessment of low-contrast sensitivity for CT systems using a model observer,” Medical Physics 38, S25S35 (2011). Hsieh, J., Computed

18、Tomography: Principles, Design, Artifacts, and Recent Advances, SPIE Press Monograph Series, 2015, chapter 5. T. Ishida, S. Tsukagoshi, K. Kondo, K. Kainuma, M. Okumura, and T. Sasaki, “Evaluation of dose efficiency index compared to receiver operating characteristics for assessing CT low-contrast p

19、erformance,” Proc. SPIE. 5368, 527533 (2004). Nyuts J, De Man B, Fessler JA, Zbinjewski W, Beekman FJ. “Modelling the physics in the iterative reconstruction for transmission computed tomography.” Phys. Med. Biol. 58 (2013) R63R96 Samei E, Richard S. “Assessment of the dose reduction potential of a

20、model-based iterative reconstruction algorithm using a task-based performance metrology.” Med Phys 2015; 42(1):314-323 Vaishnav JY, Jung WC, Popescu LM, Zeng R, and Myers KJ. “Objective assessment of image quality and dose reduction in CT iterative reconstruction,” Medical Physics 41, 071904 (2014)

21、Yu L, Leng S, Chen L, Kofler JM, Carter R, McCollough CH. “Prediction of human observer performance in a 2-alternative forced choice low-contrast detection task using channelized Hotelling observer: Impact of radiation dose and reconstruction algorithms,” Med. Phys. 40 (4), April 2013 NEMA/MITA WP 1

22、-2017 Page 13 2017 National Electrical Manufacturers Association Annex C Additional Considerations for Image Quality There are other important aspects to image quality in addition to low-contrast detectability. Some of them are listed below. Low-Contrast Detectability LCD is the ability to determine

23、 the presence or absence of a test object(s) that is similar in attenuation to its background material. Contrast-to-Noise-Ratio (CNR) CNR describing the signal amplitude relative to the ambient noise for simple and largely homogeneous objects. It depends on contrast and noise. High-Contrast Spatial

24、Resolution The high-contrast spatial resolution of a CT scanner describes the scanners ability to resolve closely placed objects. Spatial resolution is often measured in two orthogonal directions: in-plane (x-y) and cross-plane (z) (Hsieh). The MTF, related to image resolution, is the Fourier transf

25、orm of the point spread function (PSF), the systems response to a point object. The MTF and PSF are useful quantities for linear, shift-invariant systems, where the imaging systems response to an arbitrary object can be determined by convolving the true object and the point spread function (Vaishnav

26、 et al.). CT Number Linearity Linearity describes the amount to which the CT number of a material is exactly proportional to the density of this material (in Hounsfield units) with the same x-ray spectrum. CT Number Uniformity CT number uniformity dictates that for a uniform phantom, the CT number m

27、easurement should not change with the location of the selected ROI or with the phantom position relative to the iso-center of the scanner (Hsieh). Image Noise Image noise is the standard deviation of CT numbers within a ROI of a reconstructed image. Noise Power Spectrum (NPS) NPS captures the noise

28、texture. It contains the spatial frequency content of the noise in an image, and characterizes the noise correlations in Fourier space (Vaishnav et al.). The NPS has a profound impact on the LCD (e.g., Hsieh). Images with identical noise standard deviation and different NPS show different LCD perfor

29、mance. The standard deviation alone does not completely characterize the noise performance of a system (Hsieh). NATIONAL ELECTRICAL MANUFACTURERS ASSOCIATION 1300 NORTH 17TH STREET, SUITE 900 ROSSLYN. VA 22209www.NEMA.orgTO ORDER ADDITIONAL NEMA STANDARDS VISITWWW.GLOBAL.IHS.COM OR CALL 1-800-854-7179/1-303-397-79565612_0514TB