Tuesday, October 16, 2012

How to use a QC test pattern to check monitor calibration

The AAPM TG-18 QC test pattern is
essential to check monitor calibration


Whether a monitor has good image quality often is expressed in subjective terms describing quality in vague and imprecise terms. Examples of these terms are “it looks too flat,” “the resolution does not appear to be up to par,” “the brightness is not sufficient,” “I am missing contrast,” etc. 

Quantifying image quality to take the subjectivity out of the equation can be done by using an appropriate test pattern. The reason I say “appropriate” is that many users still use the “old” SMPTE test pattern, or, they might use the advanced QC test pattern such as developed by the AAPM task force and don’t know how to use this pattern, or what to look for.
If used correctly, the AAPM TG-18 QC test pattern will indicate immediately if there are any potential issues with the calibration and/or monitor. Note that in addition to using the correct test pattern, it is also critical to use the test pattern that fits your screen resolution, as there are two different files, one for high-resolution monitors and one for physician workstations.
There are significant differences between medical-grade and commercial-grade monitors, more details can be found here, one of the main differences is that medical-grade monitors have automatic correction and calibration according to the DICOM standard requirements. For commercial grade monitors it is critical that the calibration be checked visually on a regular basis, which can be done with the AAPM QC-18 test pattern.
Many users only look at the two “blocks” in the test pattern showing the 5% and 95% contrast differences, representing 5 grey scale values, however, this is only a rough indication of whether the monitor is somewhat within range, but this is by no means sufficient for medical use. This can be illustrated by the fact that, for example, a nodule in a lung field might represent a difference of 3 grey scale values. If the monitor only resolves every 5 grey scale values, you would not see a difference between those three, and the lung nodule might go undetected. That is why this test pattern shows differences that increase or decrease one grayscale value at a time starting from the black or white region in the text displayed on the bottom labeled, “QUALITY CONTROL.” In addition, there are blocks in the test pattern that have either +/- 4 digital driving levels or roughly 8 JND's. embedded at the corners.
Sometimes a vendor device handles the images improperly by applying certain look-up tables and/or image processing calculations. This can be verified as well by inserting the test pattern into the DICOM format.
Images are available for download. A link, as well as a detailed demonstration of how to use test patterns, is shown here in this video.
Each medical monitor, and commercial grade monitors which are used for medical applications in particular, should have the AAPM QC-18 test pattern available. There are TIFF versions and DICOM versions allowing you to test the complete display processing pipeline. Some institutions bring up this test pattern daily as a QA check first thing in the morning. At a minimum, I would suggest that you have users check the characteristic features of the pattern at least weekly, and, obviously, if there is any suspicion at any time that there could be an image quality problem.