Practical guidelines for color calibration and quality assurance of medical displays Poster No.: C-1140 Congress: ECR 2017 Type: Educational Exhibit Authors: T. Kimpe, J. Rostang, G. Van Hoey, A. Xthona ; Kortrijk/BE, 1 2 1 1 2 1 Beaverton, OR/US Keywords: Quality assurance, Image verification, Computer ApplicationsGeneral, Computer Applications-Detection, diagnosis, PACS, Computer applications DOI: 10.1594/ecr2017/C-1140 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 22
Learning objectives The purpose of this educational exhibit is to: 1. 2. 3. understand and experience why color calibration of medical displays is important understand what the effect is of color calibration on appearance of medical images learn practical guidelines to correctly handle color of medical displays Page 2 of 22
Background The last few years color has become more important in medical imaging. Many imaging modalities generate color images and in many cases color information is critical for obtaining a correct diagnosis. Depending on the specific field of medicine, requirements for the representation of colors may vary. For instance, for the interpretation of wound photographs, color is an indicator of the healing state of the wound, correct representation of colors is important and dermatologist are demanding standardization [1]. Other medical modalities use pseudo colors to display numerical/quantitative information on top of grayscale images. The exact color used to visualize quantitative information is less of importance as long as differences are easily perceivable and it is easy for the observer to visually determine what quantitative value is being represented by a specific color. Examples of pseudo color information are fused PET/CT images, doppler ultrasound images and contrast enhanced MRI images. Page 3 of 22
Fig. 1: Example of a PET/CT image Image source: http://www.rad.kumc.edu/nucmed/ clinical/pet_melanoma.htm References: University of Kansas Medical Center Page 4 of 22
Fig. 2: Example of a Doppler ultrasound image Image source: https:// www.healthcare.siemens.com/ultrasound/clinical-media/acuson-p300-abdominal-renalvasculature-color.html References: Siemens Healthcare Page 5 of 22
Fig. 3: Example of a contrast enhanced MRI image of the prostate. Image source: http://www.massgeneral.org/imaging/news/radrounds/august_2010/ References: Massachusetts General Hospital Department of Radiology Greyscale calibration (DICOM GSDF) [2] is not sufficient to guarantee consistent and qualitative color visualization. DICOM GSDF calibration ensures a consistent representation of greyscales, but leaves the color behavior of display systems unaltered. Up till today there is no clear standard that describes how color medical images need to be visualized. This lack of standardization results into large variability of color appearance and makes consistency and quality assurance a challenge. Page 6 of 22
Images for this section: Fig. 1: Example of a PET/CT image Image source: http://www.rad.kumc.edu/nucmed/ clinical/pet_melanoma.htm University of Kansas Medical Center Page 7 of 22
Fig. 2: Example of a Doppler ultrasound image Image source: https://www.healthcare.siemens.com/ultrasound/clinical-media/acuson-p300abdominal-renal-vasculature-color.html Siemens Healthcare Page 8 of 22
Fig. 3: Example of a contrast enhanced MRI image of the prostate. Image source: http:// www.massgeneral.org/imaging/news/radrounds/august_2010/ Massachusetts General Hospital Department of Radiology Page 9 of 22
Findings and procedure details Color variability of medical displays If not properly calibrated, display systems have a large variability in color behavior [3]. Not only will there be large differences in how individual display systems visualize color, but also on one and the same display color visualization will change over time as the display system ages [4]. Fig. 4: Example of display aging effect [4]. Original (non-aged) image is at the center with the aged versions around it as follows. From top, clockwise: Chroma aged only at 10,000 hours, chroma aged only at 18,000 hours, chroma and luminance aging 10,000 hours, and chroma and luminance aging 18,000 hours. References: Healthcare Division, Barco - Kortrijk/BE Several scientific publications have confirmed that color variability of uncalibrated displays can be very large and is highly unpredictable [5; 6]. Page 10 of 22
Potential clinical implications Inconsistent color visualization can potentially have a large clinical impact. For applications where absolute correct color is important, failing to properly handle the color behavior of the display will simply result into wrong colors being visualized. Color errors can be very large [5]. If absolute color is being used for diagnosis such as eg. in dermatology then this could create risks of wrong diagnosis. For applications where pseudo color are being visualized, failing to properly handle the color behavior of the display can result into inconsistent color visualization and unpredictable visibility of subtle color differences. In practice this means that subtle differences in color may be visible on one display system but not on another display system. Or subtle differences in color may be visible on a particular display but not anymore visible on that same display a few months later as the display ages and changes color behavior. Needless to say that this is an unacceptable situation that could influence clinical decisions, but that also could negatively influence diagnostic confidence of radiologists. Guidelines and practical recommendations The section below describes clear guidelines on how to ensure consistent color visualization. These recommendations cover both the use cases of absolute color as well as pseudo color in medical images. Absolute color For medical images where absolute color correctness is important, it is recommended to use the ICC framework [7]. ICC profiles have been succesfully used in other applications such as prepress, and have proven capable of providing accurate absolute color reproduction. Using an ICC profile as such is not sufficient to guarantee accurate color visualization. There are several types of profiles that can be used and the profile type heavily influences the accuracy of the result. More detailed recommendations on what type of ICC profile should be used and how the ICC profiles should be generated can be found in a recent whitepaper [8] published by the ICC Medical Imaging Working Group (MIWG) [9]. Page 11 of 22
Fig. 6: Summary recommendations of the ICC MIWG related to use of ICC profiles for color visualization of medical images References: International Color Consortium Moreover since each display system has its own unique color behavior, it is extremely important that an ICC profile is used that was generated for the specific individual display system being used. Generic ICC profiles that describe the 'typical' behavior of a type of display do not offer the necessary accuracy and will result into still high color errors. Some medical display systems have an integrated front sensor that allows for making automatic measurements that can be used for ICC profile generation. Advanced medical display QA software packages [10] have support for generation of ICC profiles of individual displays. Page 12 of 22
Pseudo color For medical images that include pseudo colors it is recommended to calibrate the display system such that not only the greyscale behavior is perceptually linearized but also the color behavior of the display is made perceptually linear. This will ensure that subtle differences in color will be easier to perceive and that color images look similar even on different display systems. This is especially important in quantitative imaging applications. Extensive work has been done the last few years [11; 12; 13] on this topic and a proposal has been made to extend the DICOM GSDF standard towards color in order to achieve perceptual linearity of color, while still being compliant with the DICOM GSDF standard. This proposed extension was given the name CSDF (Color Standard Display Function) and is described in a recent scientific publication [14]. This publication also suggest an easy to use visual test pattern that can be used to quickly check whether a display system is properly calibrated or not. Fig. 7: Visual test pattern to verify CSDF calibration. Display is properly calibrated if all "CSDF" letters are equally well visible for each of the colors. References: Healthcare Division, Barco - Kortrijk/BE Page 13 of 22
Formal standardization of CSDF has recently been initiated as a possible task for AAPM TG196 [15]. In view of potential future standardization some medical display manufacturers have already incorporated the proposed CSDF standard, albeit under a different (commercial) name [16] into their systems and made it possible to fully automatically calibrate medical displays by means of Quality Assurance and Calibration Software. Page 14 of 22
Images for this section: Fig. 5: JND difference per step for different colors Healthcare Division, Barco - Kortrijk/BE Page 15 of 22
Fig. 7: Visual test pattern to verify CSDF calibration. Display is properly calibrated if all "CSDF" letters are equally well visible for each of the colors. Healthcare Division, Barco - Kortrijk/BE Fig. 4: Example of display aging effect [4]. Original (non-aged) image is at the center with the aged versions around it as follows. From top, clockwise: Chroma aged only at 10,000 hours, chroma aged only at 18,000 hours, chroma and luminance aging 10,000 hours, and chroma and luminance aging 18,000 hours. Healthcare Division, Barco - Kortrijk/BE Page 16 of 22
Fig. 6: Summary recommendations of the ICC MIWG related to use of ICC profiles for color visualization of medical images International Color Consortium Page 17 of 22
Conclusion Color has become more and more important in medical imaging. Many imaging modalities generate color images and in many cases color information is critical for coming to a correct diagnosis. Greyscale calibration (DICOM GSDF) is not sufficient to guarantee consistent and qualitative color visualization. A standard that describes how color medical images need to be visualized does not yet exist. Recently a proposal has been made for a "Color Standard Display Function" (CSDF), an extension of DICOM GSDF towards color. This educational exhibit provided clear guidelines and recommendations on how to handle both absolute correct color as well as pseudo color medical images. Following these guidelines makes it possible to achieve accurate and consistent color behavior in medical images. Page 18 of 22
Personal information Disclosure: Tom Kimpe is employee of Barco Healthcare, a company developing and selling medical display systems Tom Kimpe received his masters in computer engineering from University of Ghent, Belgium in 2001. He also obtained a PhD from the same university on the topic of image quality of medical displays. In 2010 he finalized a Masters in Business Administration at the Vlerick Management School. Since 2001 Tom has been working in Barco's Healthcare Division. He has taken the positions of development engineer, project manager, innovation manager, VP of Technology and Innovation and Chief Technology Officer of Barco's Healthcare Division. Already for several years, he is coordinating all research and innovation activities of Barco's Healthcare Division and is steering an international team of researchers in USA, Belgium and Italy. The last few years Tom has been active in various standardization committees such as ICC MIWG and AAPM TG196 where he is working towards standardization of color in medical imaging. Contact information: tom.kimpe@barco.com Page 19 of 22
Images for this section: Fig. 8: Dr. Tom Kimpe, Barco Healthcare Healthcare Division, Barco - Kortrijk/BE Page 20 of 22
References [1] A. Marghoob for the International Skin Imaging Collaboration Melanoma Project Working Groups, "Standards in Dermatologic Imaging," JAMA Dermatology, vol. 151, no. 18, pp. 819-821, 2015. [2] NEMA, Digital imaging and communications in medicine (DICOM), part 14: Grayscale Standard Display Function, vol. PS 3.14, National Electrical Manufacturers Association, 2001. [3] http://www.fda.gov/medicaldevices/newsevents/workshopsconferences/ ucm342138. Last accessed December 16th 2016 [4] Avanaki, Ali RN, et al. "Aging display's effect on interpretation of digital pathology slide." SPIE Medical Imaging. International Society for Optics and Photonics, 2015. [5] Badano, Aldo, et al. "Consistency and standardization of color in medical imaging: a consensus report." Journal of digital imaging 28.1 (2015): 41-52. [6] Krupinski EA, Silverstein LD, Hashmi SF, Graham AR, Weinstein RS, Roehrig H: Observer performance using virtual pathology slides: impact of LCD color reproduction accuracy. Journal of Digital Imaging 25(6):738-743, 2012 [7] P. Green, Color Management : Understanding and Using ICC Profiles - Phil Green, Michael Kriss, P. Green and M. Kriss, Eds., Wiley, 2010, p. 314. [8] ICC, "Recommendations for visualization of medical content on color display systems", ICC_White_Paper44_visualization 2016, http://www.color.org/whitepapers/ _of_medical_color_content.pdf [9] http://www.color.org/groups/medical/medical_imaging_wg.xalter [10] QAWeb - Barco healthcare, https://www.barco.com/en/qaweb. Last accessed December 16th 2016 Page 21 of 22
[11] L. Silverstein, S. Hashmi, K. Lang and E. Krupinski, "Paradigm for achieving color-reproduction accuracy in LCDs for medical imaging," Journal of the Society for Information Display, vol. 20, pp. 53-62, January 2012. [12] Avanaki, Ali, et al. "Perceptual uniformity of commonly used color spaces." SPIE medical imaging. International Society for Optics and Photonics, 2014. [13] Kimpe, Tom, et al. "Does the choice of display system influence perception and visibility of clinically relevant features in digital pathology images?." SPIE Medical Imaging. International Society for Optics and Photonics, 2014. [14] Kimpe, Tom, et al. "Color standard display function: A proposed extension of DICOM GSDF." Medical Physics 43.9 (2016): 5009-5019. http://onlinelibrary.wiley.com/ doi/10.1118/1.4959544/full. Last accessed December 16th 2016. [15] https://aapm.org/org/structure/default.asp?committee_code=tg196. Last accessed December 16th 2016. [16] Barco Healthcare, SteadyColor calibration algorithm, https:// az877327.vo.msecnd.net/~/media/downloads/technology%20highlights/u/ 14235BARCOtechnology%20highlightSteadyColorHR%20pdf.PDF?la=en&v=2. Last accessed December 16th 2016 Page 22 of 22