Scott F. Stekel, Mayo Clinic; Heather L. Weber, RT(R)(T); Kenneth J. Stock; Kenneth A. Fetterly, PhD; Alisa I. Walz-Flannigan, PhD; Steven G. Langer, PhD

The PACS display is the final link in the diagnostic imaging process before the radiologist’s eye. It is important to ensure the display accurately renders the visual data without introducing artifacts or failing to convey the information contained in the image. Our institution uses both medical-grade grayscale liquid crystal displays (LCDs) and high-end business-class color LCDs. We have implemented acceptance testing (AT) and routine quality control (QC) procedures to ensure the correct presentation of electronic images. We will discuss our more than two years of experience in implementing these procedures.

The quality control program establishes and verifies the function and condition of the LCD for cleanliness, physical integrity, luminance response, luminance uniformity, artifacts, and color matching. At both AT and routine QC, displays are inspected for housing cracks or scratches to the display face, as well as cleaned with an optical-grade cleanser. All primary diagnostic displays have established luminance responses corresponding to the DICOM Part 14 Grayscale Standard Display Function (GSDF). This response is verified with a photometer and test patterns. We measure three to five sample points on the response curve and compare those luminance values to established control limits to verify correct functioning and conformance. If measured at AT, failing a control limit is used to eliminate unacceptable displays. For routine QC, a subjective evaluation of luminance uniformity is made with several test patterns. Both grayscale and color LCDs have a color tint. On grayscale displays, the perceived tint, expressed as a color temperature or white point coordinates, is a fixed function of the backlight and optical stack. Color displays can adjust their white point coordinates during calibration by altering responses of individual red, green, and blue sub-pixels across the range of luminance values from black to white. We have established control limits for the maximal radial distance in the 1976 CIELuv color space. A colorimeter is used on both grayscale and color displays at AT to measure the white point and compare it to those of other displays which will form a group, disallowing any displays which exceed our limits for white point coordinate differences. Additional data are also collected, including the hours of operation, which are predictive of reserve luminance left in the backlight.

Medical-grade grayscale LCDs (two and three megapixel) have been in use for over three years. These devices include built-in luminance stabilization circuitry utilizing a photometer that observes output of the backlight. Two-megapixel color displays capable of DICOM GSDF calibration have been in use for two years. Both are regularly inspected and tested to ensure correct operation as shown below:

Cleanliness issues: Not only do displays collect dust and other environmental contaminants, they are affected by pen/pencil marks, fingerprints, and food and beverage splashes. All displays are cleaned with an optical-grade solution and lint-free, microfiber cloth which is safe for the plastic materials in the display face. Some clinical areas have tried to use the same harsh disinfectants used for contaminated surfaces to clean the displays. We have found it necessary to discourage harsh chemicals and use recommended substitutes.

Luminance uniformity: Our experience has been that the LCD will not radically change its behavior unless there is catastrophic failure (such as a non-functional backlight bulb), which would be apparent to any observer in a subjective manner.

Luminance response: is a critical factor in ensuring high fidelity, consistent imaging across all the displays in the imaging fleet. At setup, we use calibration software supplied by the display vendor to install a DICOM GSDF response with fixed maximum and minimum luminance values. Our experience shows the response function (i.e., the curve shape, remains stable on all digitally-driven displays). The primary issue is display luminance output, particularly with the non-stabilized color display. The medical-grade grayscale LCD, with its integrated photometer and feedback circuitry, compensates for degradation of the backlight. The average maximum luminance value is typically within 2% of the target value. However, we have also measured gradual increases (15-20 cd/m2) in luminance output over twelve months. We speculate this may be due to decreasing sensitivity of the integrated photometer causing the system to compensate by increasing backlight intensity.

The business-class color displays have no luminance stabilization circuitry. Luminance reduction due to backlight decay is the most prevalent issue. Because the DICOM GSDF curve shape varies by luminance range, a display can fall out of conformance with the response function if the display’s overall luminance varies by too much. Compounding the problem is the significant amount of decay we have seen in backlight output over time. For example, in the first 2,000 hours of operation, average color display maximum luminance decreased an average of 10.4%. In order to maximize the overall conformance with the GSDF but minimize intervention with the display, we calibrate a display assuming a certain nominal luminance range, but then slightly increase the overall display brightness. While the conformance immediately at deployment is suboptimal, the luminance range “falls through” the optimal calibrated range over time. When the overall luminance falls below a control limit, we attempt to recalibrate the display to a higher luminance level, or replace the display. For example, we established a response function corresponding to the DICOM GSDF with a maximum luminance of 200 cd/m2 and minimum of 0.5 cd/m2. The display is calibrated and then the backlight output is increased until the maximum luminance is 230 cd/m2. Initially, this response curve is not ideal for that range, but is close. Over time, the display’s maximum luminance decays through 200 cd/m2 (at which time it is in optimal conformance) and finally 170 cd/m2 at which point our QC measurements would identify this display as requiring corrective action. We also verify the range of maximum luminance measurements of each display of a set is within control limits to prevent having one display on the high-end of acceptability and one on the low-end, but together having a discernable difference in appearance. Additional data will be presented in the poster figures.

Artifacts: These are typically from dust contamination, sub-pixel malfunction, or long-term image retention, erroneously known as “burn-in.” As of yet we have found no reasonable method to remove “burn in.”

Color uniformity: This refers to similar appearances of tinting between displays operating as a set. During setup and calibration, we ensure that displays of a set have color properties that are very close to each other. In our experience, displays of a set will maintain a similar appearance to each other for the remainder of their operational life together. While the absolute color point may change over time due to aging of the optical stack, displays will change similarly and do not cause user complaints.

The primary goals of the electronic display quality control group are to ensure the high-fidelity and accurate rendition of medical images with a consistent presentation across the entire fleet of displays. With the implementation of our QC protocols, we have been able to maintain a fleet for more than two years with efficiency and effectiveness.

The most prevalent problems found with displays are luminance output decay and display artifacts.

Samei E, et al. “Assessment of Display Performance for Medical Imaging Systems,” American Association of Physicists in Medicine Task Group 18 Imaging Informatics Subcommittee, American Assocation of Physicists in Medicine.