Are we in the dark about reading medical images?

Medical imaging specialists such as radiologists spend much of their time viewing x-ray and other images, traditionally on film but now often on computer monitors. They need to be able to spot abnormalities that show up only as subtle variations in image contrast. To make the most of the available image contrast, viewing is carried out under controlled conditions in specialist ‘reading rooms’, generally with low levels of ambient light.

Low lighting levels, however, may cause eye-strain that reduces diagnostic performance and efficiency.^1,2 This eye-strain may be because the eyes have to swiftly adapt to the bright monitor screen and then the much darker surrounding surfaces when interpreting images. Preliminary studies have concluded that consistent luminance levels between medical monitors and background surfaces can eliminate eye-strain in radiologists who have previously complained of it.^3,4

To date there has been no serious attempt to increase ambient light levels in reading rooms because of concerns that doing so will reduce the ability to spot subtle variations in image contrast, especially in darker areas of the image. Current standards for the design of reading rooms and the calibration of medical monitors therefore take little account of the problem of visual fatigue.

But with the advent of liquid crystal displays (LCDs) as the display of choice for many radiology reading rooms, a way to safely increase ambient lighting may finally be in sight. Currently available LCDs offer display properties that enable flexibility in setting ambient light levels while keeping contrast degradation under control. The methodology presented here exploits the properties of LCDs to allow reading rooms to be illuminated in a way that minimizes the luminance adaptation required of the eye—potentially reducing visual fatigue—while maximizing visible image contrast.

The approach is based on the adaptation characteristics of the human visual system and on advantageous properties of LCDs, such as low diffuse reflection coefficients and the ability to offer high maximum luminance at high luminance ratios. The hypothesis is that a discrepancy between the adaptation luminance level of the human eye when viewing an image on a display, L_adp, and the luminance values reflected off surfaces found in the reading rooms, L_s, causes notable variations in the adaptation state of the eye. This would demand continuous adjustments of the human visual system as the gaze of the reader moves from the display to the surroundings and back. The difference can be expressed as:

and minimizing its value will likely reduce visual fatigue of clinicians engaged in prolonged interpretation of medical images. The requirement to minimize f will be referred to here as ‘condition 1’.

Values of L_s depend on the diffuse reflection coefficients of the surrounding surfaces, R_s, and also, more importantly, on the ambient illumination of the room. Condition 1 therefore requires an increase in the ambient lighting level, but this in turn reduces contrast in darker portions of the image because of the resultant change in the effective luminance ratio (dynamic range) of the display.

To assure adequate contrast in the presence of ambient lighting, The American Association of Physicists in Medicine (AAPM) has made two recommendations through its Task Group 18 (TG18).⁵ First, the effective luminance ratio of any medical monitor should be greater than 250. Second, the ambient light in the room should be less than the minimum display luminance by a factor of 1.5. An optimum setup of a reading room should therefore satisfy condition 1 and the two conditions set out by TG18.

As a first step to meeting these three conditions, a computational model was developed to determine the value of the global luminance adaptation L_adp when viewing a typical medical image on an LCD. The model was based on the diameter of the pupil of the eye, which in turn depends on the luminance of the observed object.⁶

Second, the value of L_adp was compared with the luminance reflected off surrounding surfaces, L_s, under various conditions of room illumination, different values of diffuse reflection coefficients of surrounding surfaces, and calibration settings for a typical LCD.

Next, values of f for surfaces with different reflection coefficients were calculated from the values of L_s and L_adp. These are shown in Figure 1 for an LCD with a typical luminance calibration. Also shown is the change in the effective luminance ratio as a function of room illumination, and the two conditions imposed on the effective luminance ratio by the TG18 recommendations.

As Figure 1 shows, the acceptable range of effective luminance ratios set by the TG18 recommendations covers a considerable range of values of f. As a result, it is possible to meet all three conditions for ambient lighting level. In particular, room illumination in the 75–150 lux range, and surrounding surfaces with diffuse reflection coefficients in the range 0.13–0.22 sr^-1, provide an ideal setup for reading rooms with typical LCDs.

These results suggest that for typical luminance settings of current LCDs, a combination of room surfaces with higher reflectivity, and higher ambient lighting levels, can reduce visual fatigue in medical reading rooms without compromising contrast rendition in darker portions of images.