During the previous century, effective and reliable quality control measures evolved to ensure the integrity of radiographic analog film images. Over the years, these programs have achieved an equilibrium in an extremely complex technological and medicolegal environment. Radiology does not have the leisure of another 100 years to develop such measures for the rapidly unfolding digital medical environment. We need a plan now.
As outlined in his article “Reality Check: QC in the Digital Department,” John Romlein describes such a program, and while it may appear daunting when performed on top of analog film quality control, it is necessary. “It takes some special skills to develop a program, but once developed, it becomes a repeatable task, and a physicist or engineer can analyze the results,” he notes.
Leaving quality control up to whoever holds the service contract is a mistake. “You have a service man who all of a sudden is repairing a PACS workstation, and he does not understand the chain of events in the image display process,” notes Romlein. “It is not what it used to be with laser printed film. There are several intermediary devices that can have an effect on the characteristics of the displayed image. These devices include every output device in the PACS environment that has a lookup table that remaps pixel values into display luminance. This is true of laser printers, display CRTs, and display flat panel monitors.”
The ideal scenario is this: “The best way of establishing a standard display is to take a SMPTE [Society of Motion Picture and Television Engineers] test pattern that has been stored as a DICOM image and output that single image through every single display device. The result is that you have driven every display device with a standard image and you can compare the display performance across the gray scale range of all display devices in the system.”
What is more common, however, is this: “Most systems have a test pattern embedded in them, but it is a local pattern either within them or in the calibration software. The problem is that it is a local image and the exact same test pattern is not necessarily stored in all devices. It is not apples to apples: you are comparing local output to local output. Even if you equalize all of the local display characteristics, there is nothing to say that when a global image from PACS is outputted through these devices, it will have the same characteristics. One of the reasons is the separation of the calibration process of local devices from their display software. You may well have another processing step that will kick in and create differences in the displayed images.”
No one has created a commercial standard DICOM test pattern product that can be stored in the PACS, although there is reportedly a set of test patterns available as freeware, but the source remained elusive at press time. Romlein identified several other issues of concern:
- Drift. If monitor and calibration checks are not performed frequently enough, then monitor brightness and contrast drift create a wide disparity in the quality of displayed images. The normal recommended frequency for calibration is quarterly. It should really be done monthly. (Some new CRT and flat panel technology has built-in calibration circuitry that senses drift.)
- Black level. Most monitors and calibration test equipment are good at controlling the maximum luminance of the display but do an inadequate job of controlling the black level.
- Image scaling. There are known magnification factors used by orthopedic surgeons to match implant templates to a radiographic image for the purpose of sizing the implant. The PACS world has not adequately addressed magnification quality control when images are displayed on monitors and laser printed film, though some vendors have.
Our is a transitional time in radiology. Behind us lies the analog world, and before us lies the digital world with all of its operational efficiencies and clinical promise. Then there is the world in which most of us exist today, with its combination of analog and digital modalities and operations. It is not unlike the Purgatory described in Dante Alighieri’s The Divine Comedy: There are no angels with harps, there are no 100 virgins, and there is no early retirement. However, you’ll find most of your colleagues there. n