Evaluation of Radiologist Preferences for Alternative Colors
for Monochrome Display of Images in Conventional Radiology and CT
 
Authors:
Frederick E. Weiss, MD, DPT, University of Maryland Medical Center; Joseph J. Chen, MD; Naomi J. Saenz, MD; Alex Natanzon, MD; Emil Georgiev, MD; Eliot L. Siegel, MD, FSIIM
 
Background:
Despite the fact that film used in radiology is almost always tinted (for example with a blue dye), the transition to filmless radiology using first generation high resolution grayscale computer monitors limited radiologists and clinicians to the use of various shades of gray, eliminating this tint and its aesthetic and diagnostic advantages. Blue tint was first added to film in 1933 in order to make it “easier to look at.”[1] Unfortunately, the transition to high luminosity, high resolution, active matrix color LCD monitors has not been associated with a re-exploration/introduction of the added value of a colored background tint or the use of a color monochrome or even multiple color display of images such as conventional radiographs or CT. Consequently, radiology images are almost exclusively viewed using a grayscale lookup table on current PAC systems/imaging workstations, despite the fact that film is still manufactured with a slight tint.

Psychoperceptual studies show that humans can perceive improved differentiation in luminosity (differentiate shades of the same color) in other colors, especially green and yellow-green. Our study evaluated individual preferences for images displayed in different colors to provide baseline data for a study on the utility of monitors using other colors for monochrome and potentially even multi-color image display for conventional radiography, CT. and other imaging studies.
 
Evaluation:
Prototype software was developed using a GE PAC-IW workstation. which was designed to display conventional radiology and CT images (as well as for other modalities) in monochrome. using various portions of the visible spectrum of light, in addition to the standard grayscale display. The multiple color schemes were rendered by adjusting the HSV (hue, saturation, value) and RGB (red, green, blue) values. Light blue and light yellow-green were selected for comparison with the grayscale display. Five digital images (AP chest and AP pelvic radiographs, thoracic and abdominal CT images, and a left MLO mammogram) were selected and de-identified . Eight readers were recruited to look at each digital image in two color schemes as well as using the conventional grayscale display, and to rate the diagnostic properties of them in order of preference. These color schemes were light blue and light yellow-green.

Overall, the light yellow-green display was selected most often as the optimal display frequency (43.2%) for the 5 different digital images, followed by grayscale (36.4%) and light blue (20.5%), mostly due to a very strong preference for pelvic radiography. Interestingly, preferences varied based on imaging modality and anatomic area examined. For chest radiography and abdominal CT, grayscale was preferred by 43.8% and 41.7%, respectively, followed by light yellow-green (34.4% and 33.3%) and light blue (21.9% and 25%). For pelvic radiography, light yellow-green was preferred (50%), followed by light blue (29.2%) and grayscale (20.8%).
 
Discussion:
The eye sensitivity function for photopic vision (1 candela/meter2 through 1 million candela/meter2) suggests that luminous efficacy is highest at 555 nm, which falls within the higher wavelength range of the green portion of the visible spectrum near the yellow spectrum.[2]

This work was originally performed by Gibson and Tyndall[3] and strongly suggests that our eyes have the ability to perceive more differentiation in luminosity in the green and green-yellow portion of the visible spectrum than in any other. Potential benefits of switching to monochromatic green (507 nm) or yellow green (555 nm) could include improved reading efficiency and minimized eye figiture. Other monochromatic image viewing technologies, such as night vision devices[4] and modern aircraft displays, utilize monochrome green at approximately 507 nm for their displays.

There are a number of important limitations of our study which are also opportunities for future research. Our readers were inexperienced with clinical image interpretation at those different wavelengths, and this novelty was very likely an important consideration in their determination of which of the three color choices was optimal. Additionally, we did not test the speed or accuracy of the readers in rendering a diagnosis. The preferences of the readers may not have correlated well with the colors that allowed them to perform optimally from a diagnostic or efficiency perspective.

Additionally, the monitors utilized in the study kept backlight luminosity constant, but the overall output of the monitors may have varied with different colors. It would, for example, have been interesting to measure how luminosity changed for a chest radiograph with changes in color on a given monitor. Finally, it would also be very useful to determine the impact of variation in background color, keeping the foreground as grayscale information but modifying the background color to better simulate the look of tinted film.
 
Conclusion:
There were significant variations in color preference among 8 readers for conventional radiography of the chest, pelvis and for body CT studies. The yellow-green display were overall preferred over the others despite participants’ lack of previous experience with color although preferences varied considerably according to anatomic location and modality. Our findings are consistent with the psychophysiological work performed almost 90 years ago but not implemented by any of the current image display vendors. Additional studies should be performed that control for absolute image luminosity and test for diagnostic accuracy, performance, and/or confidence.
 
References:
1. Murry RC, Dowdey JE, Chirstensen EE. Christensen’s physics of diagnostic radiology, 4th ed. 1990;138.

2. TG18 Report, Med. Phys. April 2005;32(4).

3. Gibson KS, Tyndall EPT. Visibility of Radiant Energy, Scientific Papers of the Bureau of Standards. 1923;19 131-91.

4. http://www.hownightvisionworks.com