Table of contents
- What is ambient light?
- About Just Noticeable Differences and Digital Driving Levels
- Why is ambient light compensation so important when viewing medical images?
- How do we compensate for ambient light when DICOM calibrating a display?
- Further reading
Definition: Ambient light is the light which is present in an environment.
In a diagnostic reading room context: Ambient light is the surrounding light in a reading room which can influence diagnostic reading:
- Reflection on the display
- Light from the environment shining into the eyes of the viewer (e.g. a light box next to the diagnostic display)
Assume an 8-bit grayscale image. This image can consist of 256 different shades of gray. Each of the 256 shades of gray is called a Digital Driving Level (DDL). For each pixel, the display controller sends out a pixel value between 0 and 255 which is then converted by the display into visible light.
The intention of the DICOM calibration is to have a visual consistency on how a given digital image appears. DICOM takes into account the contrast sensitivity of the human eye. In practice, this means that consecutive digital driving levels correspond with an equally visible luminance difference. That visible luminance difference between two subsequent digital driving levels consists of Just Noticeable Differences. Only this way we can be sure that a low-contrast pathology (such as for instance a lung nodule) is going to be easily discernable by the radiologist.
Without going too much into detail on DICOM, the images below should give you a better visual idea of what DICOM really does:
Without DICOM calibration (native curve):
With DICOM calibration:
Ambient light which is present in a diagnostic reading room (and any other environment of course) reflects onto surfaces back into our eyes. Because an LCD screen is also a surface, ambient light will reflect on your display's screen surface as well. This will result in an overall contrast loss.
Walk outside on a sunny day with your phone or tablet. When your phone or tablet screen is exposed to the bright sunlight, it will be very hard to read.
This contrast loss is most critical in the dark areas of the screen because it may actually make critical information in a medical image a lot less visible. In many cases, very small luminance differences in the darker areas could indicate the presence of a certain pathology (e.g. a lung nodule). The DICOM calibration aims to make these contrast differences as visible as possible. This works well if correctly applied.
Weber's law states that our senses are most sensitive to relative changes, changes relative to the starting point. Knowing that the following example should bring more clarity (this is a strong simplification of the actual situation and calculations):
In a dark room:
- Ambient light = 0 Lux
- Reflected ambient light on the screen is 0 cd/m2
- DDL0 (black point) is 0.30 cd/m2, DDL1 (next step) is 0.35 cd/m2. The relative luminance difference between DDL0 and DDL1 is 17% (Formula: (0.35/0.30)*100)
In a room with ambient light:
- Ambient light: 100 Lux
- Reflected ambient light on the screen is 0.6 cd/m2
- DDL0 (black point + ambient light) is 0.90 cd/m2, DDL1 (next step) is 0.95 cd/m2. The relative luminance difference between DDL0 and DDL1 is 5.5% (Formula: (0.95/0.90)*100)
In this example and without compensation for the available ambient light, the difference between DDL0 and DDL1 will be a factor 3 less visible. If you know that our eye is very sensitive to differences, then it is easy to understand that in the latter situation, the room for diagnostic error is much higher.
To avoid this, an ambient light compensation can be applied to a display. Keep reading to find out how this ambient light compensation is calculated and applied.
The ambient light reflected by the display can be calculated with the following formula:
Lamb = Rd * Illuminance
- Lamb = Reflected ambient light (in cd/m2)
- Rd = The reflection coefficient of the display (an individual coefficient per display type, stored inside each Barco diagnostic and clinical display)
- Illuminance = The average ambient light in the room, measured with a lux meter (in Lux)
Here is an example of an actual Lamb calculation:
- Display type: MDNC-3421 (Nio Color 3MP)
- Reflection coefficient: 0.00803 (this information is read out from the display by QAWeb Agent during calibration, no need to know this)
- Measured ambient light: 35 Lux
- Lamb = 0.00803 * 35 = 0.028105 cd/m2 (note that QAWeb Agent will show this value as a rounded value)
Note: Alternatively the Lamb can also be measured by turning off the display and measuring with a distance luminance meter about 50cm from the screen surface, perpendicular to the screen. This method is more difficult, especially when there is low ambient light. Most sensors cannot measure the low Lamb value accurately.
As the reflected ambient light influences the diagnostic reading, it should be compensated for during the calibration of the display system.
For calibration, ambient illuminance is not measured, but rather selected, i.e.as part of one of the reading rooms described in the QAWeb User Manual; this illuminance is used to calculate an Lamb per the formula above. During calibration, throughout the entire dynamic range of the display, each DDL gets appointed a certain ideal luminance (this includes this selected ambient luminance Lamb). This luminance is based on table B-1 in the NEMA DICOM Part 14 document and the luminance range is between the black luminance (+ Lamb) and the white luminance (+ Lamb). The end result is an equal amount of JND's between each DDL.
Going back to our example:
Without ambient light:
- DDL0 = 0.3 cd/m2
- DDL255 = 400 cd/m2 (assuming this is the calibrated brightness of the display in our example)
- The dynamic range of this display is 400/0.3=1333
- With the NEMA JND conversion table: 639 JND's/256 steps ~ 2.49 JND's difference between two DDL
With ambient light:
- DDL0 = 0.3 cd/m2 + 0.6 cd/m2 = 0.9 cd/m2
- DDL255 = 400.6 cd/m2
- The dynamic range of this display is 400.6/0.9 = 444
- With the NEMA JND conversion table: 605 JND's/256 steps ~ 2.36 JND's difference between two DDL
So even though the ambient light in the room is pretty high and the reflected ambient light quite is high as well, the DICOM calibration with ambient light compensation will keep a good level of JND differences between each DDL. If you compare that with the example in the previous paragraph where the visible contrast between two DDLs was a factor 3 less (without proper ambient light compensation), you see the immediate importance of proper DICOM calibration with ambient light compensation.
- A reference table which shows each JND step mapped to an actual luminance can be found in the NEMA DICOM Part 14 reference document.
- The theory between DICOM's perceptual linearization is based on the work by P. Barten, which is a more sophisticated model of human contrast sensitivity than Weber's law.
- Everything you want to know about DICOM can be found here: https://www.dicomstandard.org/
Last updated Feb 03 2020