The impact of computer display performance on the quality of digital radiographs: a review
Dr Alison Butt
Brisbane Dental Hospital
Cnr Albert and Turbot St
Brisbane QLD 4000
Radiography makes an essential contribution to the processes of examination, diagnosis and treatment planning in dentistry. While the use of film-based imaging still predominates in dentistry, digital imaging is gaining wider acceptance and the use of this modality is anticipated to expand in the future. Two concerns associated with this transition have been raised in the literature. The first of these is the dissatisfaction experienced by many dental professionals with quality of digital radiographs when compared to plain film. In addition, there are indications that practitioners feel limited in their understanding of those factors impacting on digital image quality.
One key area highlighted in the literature as having a significant influence on digital radiographic quality and interpretation concerns the performance of the display device. Within the last decade, research derived from the fields of medical radiology and physics have demonstrated that suboptimally performing displays degrade image quality, thereby increasing the potential for compromised diagnostic outcomes. In the field of medicine, this has resulted in the establishment of standards applicable to computer displays used in diagnostic radiology. Conversely, limited guidelines exist in the field of dentistry. The aim of this review is to provide an outline of these standards and highlight the important relationship between computer display performance and digital image quality.
Abbreviations and acronyms:
American Association of Physicists in Medicine
- AAPM TG18
American Association of Physicists in Medicine Task Group 18
American College of Radiology
active matrix liquid crystal displays
Digital Imaging and Communications in Medicine
US Food and Drug Administration
International Standards Organization
UK Medicines and Healthcare Products Regulatory Agency
US National Electrical Manufacturers Association
pixels per inch
Society of Motion Picture and Television Engineers
Concept of the digital imaging chain
The generation of a radiographic image, whether analogue or digital, is consequent to a series of steps. In analogue radiography information is recorded on emulsion-coated film, which then undergoes chemical processing to reveal the latent image. Techniques concerning patient positioning, exposure parameters and image processing all may influence the quality of the final product. Digital imaging is also dependent on these factors with the exception that information collection and processing is undertaken through electronic means.1 The distinct separation of the process of image acquisition, processing and display in digital radiography presents both advantages and disadvantages. With analogue film, alteration of image quality is possible at the time of image acquisition and processing. However, little or no manipulation of the image is possible after chemical processing has been completed. In digital radiography, the opportunity exists to influence and manipulate information at any time after image acquisition. A digital image can be altered as a consequence of the type of computer hardware, software and graphics system employed, or the type of monitor utilized to display the image. The capacity to alter digital images post-processing is also possible, and offers the opportunity to enhance image quality. Conversely, when digital image quality is less than ideal, it can be difficult to determine the precise cause.1,2
The aim of any radiography system should be to present images of optimal quality. In transitioning to digital radiography, resultant image quality should at least parallel that of plain film. Many factors influence the quality of digital radiographs, and to place these in context several authors have coined the term ‘digital imaging chain’.1–4 Sorantin5 outlined that the stages involved in digital image acquisition include: obtaining the raw data, data processing, image display and interpretation of the image by the observer. This sequence of events occurs within an environment which must also be considered part of this ‘chain’.6
The final quality of a digital radiograph is dependent on how electronic information is managed through the entire chain.7 Degradation at any point results in loss of electronic information by the system which can adversely affect image quality and diagnostic outcomes.5 An extensive review into each aspect of the ‘digital imaging chain’ is beyond the scope of this review. However, it is suggested that one important factor impacting on the quality of digital radiographic images relates to the operating characteristics of display devices used to view digital radiographs.
Types of display devices
Display devices used in medical radiology are classified according to both physical type and the purpose for which they are utilized. Classification of display type essentially describes the technology employed by the device to produce light. Displays present images through complex processes which begin with the collection of information gathered during image acquisition. This information undergoes conversion by the graphics or video card into electronic signals which subsequently controls the amount of light generated by the display required for image production.4 It is important to appreciate that the entire system that drives the display is as critical to image quality as the display itself.8,9
Many types of display devices are commercially available and newer technologies are continually emerging. The most commonly utilized display currently employed in medical radiology is the active matrix liquid crystal displays (AMLCD).8 AMLCD displays produce images by electronically controlling the transmission of light through a liquid crystal layer.10 The application of an electrical field through a transistor (also referred to as a thin-film transistor) affects the orientation of the liquid crystals in each pixel or liquid crystal ‘cell’, relative to the front and rear polarizing filters. The variable orientation of these crystals controls the orientation of the polarizing filters and the transmission of light, enabling image generation to occur.8 In recent times newer technologies, in particular LED displays, have become commercially available. Unfortunately, no research concerning their operational performance relative to diagnostic radiology has been conducted. The majority of research and quality control guidelines in recent times are limited to the performance of AMLCD displays.
Display devices are also classified according to the purpose for which they are utilized. This is unique to displays used in the field of medical imaging and is recognized by the US Food and Drug Administration (FDA), the International Standards Organization (ISO) and the Medicines and Healthcare Products Regulatory Agency (MHRA) in the UK. Display devices for medical imaging are classified as either primary or secondary grade devices. Primary displays refer to those devices used for the interpretation of medical images upon which a diagnosis is made, and as such are expected to perform to the highest standards. Secondary displays refer to devices used to view medical images for purposes other than providing a medical diagnosis.
Display performance standards applicable to diagnostic medical radiology
Several organizations have developed standards applicable to displays used in diagnostic radiology. The primary purpose of such guidelines is to ensure display devices perform consistently within acceptable limits, and adjustment or replacement can occur when necessary. The formulation of such guidelines has been derived from a combination of expert consensus11 and research demonstrating adverse degradation of image quality consequent to suboptimally performing displays.3,7–9,12–16
The Society of Motion Picture and Television Engineers (SMPTE) was one of the first organizations to develop guidelines from 1980 onwards and ‘SMPTE RP 133-1991’17 was the standard most commonly referred to in the assessment of medical display quality. In more recent times, an expert consensus subcommittee of the American Association of Physicists in Medicine (AAPM) and the National Electrical Manufacturers Association (NEMA) developed two comprehensive guidelines – The Assessment of Display Performance for Medical Imaging Systems11 and the Digital Imaging and Communications in Medicine (DICOM) Part 14 Greyscale Standard Display Function standard18– commonly referred to as ‘DICOM Part 14 GSDF’.
Recognition of the importance of displays in the practice of diagnostic medical radiology3,8,13–15,19 and the rapid growth of digital imaging in healthcare13,20,21 has largely motivated the development of these comprehensive guidelines. Although these standards have not been legally mandated in any country (with the exception of Germany22), much of the research in diagnostic medical radiology is fundamentally based on compliance with these standards.
Display performance factors impacting on the quality of digital radiographs
It is important to appreciate that suboptimally performing displays, or the incorrect choice of a display appropriate for the diagnostic task, may compromise image quality.23 However, the display can only be considered in the context of the entire imaging ‘chain’. High performance displays complying with all known standards will offer very little benefit if the equipment used to capture and process the information is substandard and the reverse is also true. Additionally, even if the equipment is of the highest quality and performing optimally, a less than ideal viewing environment can contribute to image degradation. Displays, as such, cannot be considered independently from all other factors that contribute to final image quality. However, this review will focus on those factors directly related to the display device.
Applying standards for display performance is necessary for two reasons. Firstly, as highlighted by Badano in 2004,8 displays always degrade information contained in an image. Therefore, it is vital to ensure that certain measures are taken to limit this loss as much as possible.12 Secondly, it has been recognized that the diagnostic performance of displays always lessens with time11 and often occurs with increasing age and continuous use. With AMLCD displays this is most frequently related to a reduction in luminance (brightness) due to a decline in performance of the fluorescent backlights.11,24,25
The most important difference separating medical grade monitors from standard or commercially available displays relates to compliance with the recommendations developed by the AAPM TG1811 and NEMA.18 Compliance to these standards ensures that the display device does not unduly degrade digital image quality. A summary of these parameters, and the impact on digital image quality, are outlined as follows.
Digital Imaging and Communications in Medicine (DICOM) Part 14 Greyscale Standard Display Function
In medical radiology, compliance to DICOM Part 14 GSDF standards18 is considered critical and represents the most important difference separating medical grade displays from standard or commercially available devices. The purpose of this standard is twofold. Firstly, it was developed to ensure the consistent presentation of greyscale (and therefore contrast) in images across a wide range of displays without significant variation or degradation. Secondly, this standard ensures that the presentation of greyscale by displays is matched as closely as possible to the contrast sensitivity of the human visual system.11
The concept of the DICOM Part 14 GSDF may be appreciated by imagining a single image presented on several different display devices driven by the same computer system. For an image to be presented across different displays with little perceivable difference in image quality, the emission of light from each display must be controlled uniformly. The DICOM Part 14 GSDF standard achieves this by assigning a consistent ‘brightness’ or ‘luminance’ value to a particular input electronic signal generated by the computer graphics card. Additionally, the displayed luminance values are simultaneously matched to the contrast sensitivity of the human visual system, based on the model developed by Barten.26 The purpose of the DICOM Part 14 GSDF is to standardize the emission of light from displays, thereby ensuring the optimal presentation and perception of contrast in digital radiographs.
Display performance criteria according to the American Association of Physicists in Medicine Task Group 18
Geometric distortion refers to any alteration of digital images by the display device.11 Almost all standard and medical grade AMLCD devices will pass this criterion because display architecture is composed of a fixed arrangement of pixels. However, geometric distortion can be induced in presented images if the incorrect display resolution is chosen.11 Therefore, it is important to ensure that displays are adjusted optimally to meet this parameter.
The reflection characteristics of displays encompass both specular and diffuse phenomena. Specular reflection occurs when the display face behaves like a mirror, reflecting incident light directly back to the observer. Diffuse reflection, sometimes referred to as ‘haze’, describes the internal scattering of incident light that is reflected diffusely from the display surface.11 Excessive display reflection interferes with the perception of image contrast by the observer14,27 and is frequently the result of using displays exhibiting glossy finishes or viewing digital radiographs in bright room lighting.9 Consequently, when choosing displays preference should be given to devices demonstrating a matt finish. In addition, when viewing radiographs ambient room lighting should be reduced to minimize the reflection of light from the display face.
One of the critical aspects of display performance impacting on image quality concerns the luminance response characteristics of the device. Luminance response comprises several criteria and defines how a display controls the emission of light critical for image generation. The standards applicable to this parameter have been developed through the combined efforts of the American Association of Physicists in Medicine Task Group 18 (AAPM TG18),11 NEMA18 and the American College of Radiology (ACR). The AAPM TG1811 defines the luminance response of displays as comprising three fundamental criteria – maximum and minimum luminance (L’max and L’min), contrast or luminance ratio (L’max/L’min) and contrast response (κδ). Maximum and minimum luminance refers to the maximum and minimum amount of light emitted from the display.28 The difference between these two values defines the contrast ratio, more commonly referred to as the ‘dynamic range’ of a display. This is often expressed as a ratio to 1 and defines the limits of light a display can emit, from the very darkest to the brightest. The control of these parameters is influenced by several factors including the input signals derived from the graphics card, the physical capabilities of the display related to backlight function and the quality of the transistors conveying the electrical currents, which control the behaviour of the liquid crystal molecules.8
Displays demonstrating larger contrast ratios may potentially present a greater number of ‘shades of grey’ or contrast in a radiographic image in comparison to a display with a lower contrast ratio.10 However, a high contrast display will not necessarily present all possible ‘shades of grey’ or contrast within an image with accuracy. The consistent presentation of all contrast within the dynamic range limits of a display is defined by the final characteristic of contrast response (κδ). In simplified terms, contrast response describes the capability of a display to reliably and accurately present greyscale in response to a specified input signal.11 The contrast response characteristics of displays are established in medical systems through a process of display calibration to meet the standards developed by the AAPM TG1811 and NEMA.18
All modern day medical display devices are manufactured to meet these standards. Additionally, medical grade devices can present luminance values (light) at the very extreme ends of the luminance spectrum (L’min and L’max) with greater accuracy and stability. By expanding the range within which light may be emitted, larger contrast ratio values are achieved.8,11 Although many commercial displays demonstrate large contrast ratios, it is important to appreciate that the presentation of greyscale within that range is inconsistent because they do not conform to DICOM Part 3.14 GSDF standards. Therefore, optimally performing medical displays will present greyscale values with greater accuracy in comparison to standard ‘off the shelf’ displays that have not been manufactured or calibrated to meet these standards.
This defines the uniformity of light emission across the face of a display device.11 Significant variation in the emission of light across the display adversely affects the luminance response characteristics of a display, which in turn will affect the contrast of displayed images. The major factors contributing to luminance uniformity in AMLCD displays include the quality of the pixel architecture, uniformity of the liquid crystal layer and degeneration of the fluorescent backlights consequent to ageing.11
Display resolution commonly refers to the number of pixels in the vertical and horizontal dimension of a display.5 However, resolution in the truest sense refers to the capacity of a device to present detail in an image as separate resolvable points.11,29 One important factor influencing resolution concerns the consistency of a display to accurately present contrast and this is determined by the luminance response characteristics of the device. The second factor influencing display resolution is pixel size, more commonly referred to as pixels per inch or PPI8,10 and pixel pitch, representing the physical spacing between the individual pixels. Pixels represent the smallest picture element of the display through which light is emitted. In AMLCD displays, the physical pixel substructure determines both pixel size and pixel pitch.8 Images presented on displays with smaller more closely spaced pixel elements appear sharper with superior resolution compared to displays with larger and more widely spaced elements.29 Physically larger display devices do not necessarily provide improved image quality. As screen size increases, so do the number of pixels; however the size of the individual pixel elements may also increase.4 Lower quality displays usually demonstrate larger and more widely spaced pixel elements because they are cheaper to manufacture, whereas displays with smaller picture elements are usually associated with high quality photographic and medical display devices.8
An added factor contributing to display image quality is noise. This refers to any unwanted signals that interfere with true image detail.12 Noise can be of considerable concern because it creates difficulties in the perception of subtle changes in contrast within an image.27 Many factors contribute to noise, some of which are directly related to the display itself. The uniformity of the liquid crystal layer and display pixel architecture are both important contributors of noise in AMLCD displays.11 Peripheral to the display, the quality of the electrical signals transmitted to the display can represent a major source of image noise. An important but often forgotten contributor is the type of cable used to connect the computer system to the display. Preference should be given to using digital visual interface dual link or DVI-D cables instead of video graphics array or VGA cables to reduce electronic noise as much as possible. All currently manufactured medical grade displays utilize DVI-D connections.
Additional factors impacting on digital image quality
The display driver or graphics card
One critical factor independent of the display is the quality of the graphics card. Graphics cards convert an electronic signal into a digital driving level that is then used to generate an image. Cards of poor quality are physically unable to support and present adequate amounts of information, resulting in information loss and inferior quality images.4,5,8,24 Compared to high quality graphics cards, medical grade cards are manufactured to meet the AAPM TG18 and DICOM Part 14 GSDF standards for display contrast and resolution. In addition, appropriately matched graphics cards are always recommended and often included in the sale of medical grade displays, emphasizing the importance of these devices to display performance.
Display performance and digital image quality
The relationship between display performance on digital image quality has been examined in several papers. Of interest are studies comparing the performance of commercial grade displays with medical grade devices.28,30 Medical grade monitors are expensive and the replacement of medical grade devices with commercial displays could translate to considerable cost savings. However, both Krupinski et al.31 and Sim et al.32 were unable to optimally calibrate commercial displays to both DICOM and AAPM standards. The displays were therefore deemed inadequate for the purposes of diagnostic medical radiology.
Of considerable importance is research evaluating the relationship between suboptimal display performance and adverse diagnostic outcomes. Unfortunately, these are limited and may in part be explained by the fact that the AAPM TG18 recommendations are relatively new. The study by Buls et al. in 200716 investigated the relationship between display performance and the ability of experienced medical radiologists to detect simulated lung nodules on radiographs. At the commencement of the study, five displays were evaluated utilizing the acceptance criteria developed by the AAPM TG1811 and NEMA. The performance characteristics of each display were recorded but the devices were not adjusted if they were found to be performing suboptimally. This is of considerable importance as most studies assessing display performance have focused on devices that were optimally adjusted prior to commencing the study. After initially evaluating all displays, the authors discovered that all, with the exception of one, failed between one and several of the AAPM TG18 acceptance criteria. When a correlation between the diagnostic performance of the radiologists and the performance of displays was made, a statistically significant reduction in observer performance was observed with those displays, either demonstrating smaller contrast ratio values or maximum luminance levels below the AAPM TG18 standards.
Research studies specific to the dental profession have also investigated display performance in the review of digital dental radiographs. Cederberg et al. in 199933 investigated observer variation in the ability to detect artificial enamel lesions using four different types of AMLCD displays. The authors concluded there were no differences in diagnostic performance between the display types. However, two concerns regarding this study must be highlighted. Firstly, no mention was made of displays being adjusted for contrast or brightness using a suitable test pattern or whether display luminance values beyond the specifications provided by the manufacturer had been ascertained. A review of this paper suggests that the displays tested exhibited vastly different maximum luminance values, a factor well recognized as having a significant impact on image quality and diagnostic outcomes.16 Another concern is the use of artificially generated lesions which, as demonstrated in the study by Kang et al.,34 are 1.4 times more likely to be detected in radiographs compared to natural lesions. However, in their conclusion, Cederberg et al.33 recognized the above limitations and suggested further study utilizing examples of natural carious lesions.
Ludlow et al.35 compared the performance of film, standard CRT display monitors and AMLCD display laptops in caries detection, and also concluded there were no significant differences between the display types. However, the authors did not reduce room lighting when reading digital images. Potential reflection from the face of the CRT display associated with increased ambient lighting reduced the observers’ perception of image contrast,36 and this introduces a significant error into the study. In addition, no mention was made as to whether any of the displays were adjusted for brightness or contrast, or whether any displays tested conformed to any specific standards for display performance. In retrospect, it is apparent that the above studies did not consider several critical factors, which may have adversely impacted on the performance of the devices tested. However, these errors are understandable considering the above studies were undertaken prior to the release of comprehensive standards guiding display performance.11,18
More recently, Hellén-Halme et al.9 investigated several key aspects of quality in digital radiography. In a component of their field study, images obtained from low and high contrast test phantoms were presented on the display systems of 19 dental practitioners both prior to adjustment for contrast and brightness, and after visual evaluation and adjustment using the TG18-QC and TG18-CT test patterns developed by the AAPM TG18.11 The authors observed that after displays were adjusted, a noticeable improvement in the perception of image contrast and detail occurred. This highlights the potential benefits which may be realized if display brightness and contrast are adjusted using test patterns designed for such purposes. However, manually adjusting the brightness and contrast of a display does not ensure compliance to the DICOM Part 3.14 or AAPM TG18 standards. In displays not manufactured to these standards, display calibration requires the application of dedicated software and testing tools to adjust both the graphics card and the display, to optimize the luminance response characteristics of the device.
More studies comparing the diagnostic differences between high quality medical displays and standard displays have been undertaken. When Hellén-Halme et al.37 compared the performance of a standard desktop monitor with two DICOM Part 14 GSDF compliant medical grade displays in the diagnosis of proximal surface caries, no appreciable difference between the commercial and medical grade displays was observed. The standard desktop displays had been adjusted visually for brightness and contrast using the TG18-QC and TG18-CT test patterns and the medical grade monitors had been adjusted using an ‘inbuilt’ DICOM calibration software. Although no differences in diagnostic outcomes between the display types were identified, concerns with this study should be raised. Although it was cited that the medical displays tested complied with DICOM standards, no reference was made to compliance with any other standards, including those suggested by the AAPM TG18.11 Upon closer examination, two of the three displays tested exhibited maximum luminance levels well below that recommended by the AAPM TG18 for primary grade displays. This is of some concern because a reduction in the maximum luminance can significantly reduce the contrast ratio values achieved by the display.16 In addition, reliance on ‘inbuilt’ calibration devices should be approached with some caution. Display performance using inbuilt calibration devices should be confirmed using the methods recommended by the AAPM TG18.11
In the study undertaken by Isidor et al.38 differences in the performance of medical grade and standard displays in the diagnosis of natural proximal surface lesions were also examined. Although the authors observed that one of the medical grade displays demonstrated significantly greater accuracy compared to two of the older displays tested, no significant difference in performance between the newer medical and standard office grade displays was found.
Several concerns with this study must be raised. Firstly, the authors stated that all medical grade devices had been calibrated at the time of installation. However, there was no indication that the medical displays had been calibrated immediately prior to the study or assessed for compliance with the AAPM TG18 standards. Secondly, the authors relied on inbuilt calibration devices to ensure compliance to DICOM Part 14 GSDF standards. However, this should be verified independently using the protocols developed by the APPM TG18. Variation in the age of displays tested, and possible non-compliance with the parameters outlined by the AAPM TG18, may have adversely impacted on display performance.11,24,25 Review and calibration of the medical grade monitors against recommended standards prior to the study may have yielded different results.
A great deal of research in the field of digital radiology has been undertaken in the last decade, which highlights the growing role this technology will play in future healthcare. Several perceived benefits in transitioning from plain film to digital imaging include lower radiation dosages to the patient, the ability to link and store images within an electronic health record and the elimination of film and dark room processing procedures. Considering that the acquisition and interpretation of a digital radiograph is consequent to a series of steps, the quality of the final product is dependent on what occurs during each stage of this process.
Expert consensus from the fields of medical radiology and physics emphasize the important relationship between display monitor performance and digital image quality. Consequently, this has important implications on the practice of digital radiography in dentistry. Although research in the field of dentistry has not demonstrated conclusive proof of the advantages of employing medical grade displays over standard display devices, some of these studies demonstrate significant limitations.
The use of digital imaging in dentistry will continue to grow in the future. Therefore, the development of clear guidelines, education and research specific to this technology should be given priority.