Journal of the

A preview of the papers appearing in the October 2006 issue of the Journal of the SID. To obtain access to these articles on-line, please go to

Edited by Aris Silzars

MTF measurement method for medical displays by using a bar-pattern image

Katsuhiro Ichikawa
Yoshie Kodera
Hiroshi Fujita

Kanazawa University, Japan

Abstract — A modulation-transfer-function (MTF) measurement method that uses a bar-pattern image for medical displays such as liquid-crystal displays (LCDs) and cathode-ray tubes (CRTs) has been investigated. A specific bar-pattern image on the display was acquired with a high-resolution single-lens reflex-type digital camera equipped with a close-up lens. The MTF was calculated from the amplitudes of the fundamental-frequency components, which were extracted from the profile data across the bar patterns by using Fourier analysis. Actual comparisons with the conventional line technique were performed for a medical CRT. The adequate accuracy and excellent reproducibility of the method were confirmed. Furthermore, unlike the line method, an advantageous feature which can use an input signal with sufficient amplitude was theoretically proved. Horizontal and vertical MTFs at the central position of the display area were measured up to the Nyquist frequency for several medical displays. From these measurements, this method has the capability to detect slight differences between the displays measured. This proposed method is useful in understanding and quantifying the medical display's performance due to excellent reproducibility and accuracy.

For recent medical-image diagnosis, a diagnostic style using a softcopy display is expanding to medical fields. This trend was brought about by the rapid progress of technologies that include medical digital imaging modality, image communications, display, etc. Many arguments have been made about the image quality of the display, which is the final media for diagnostic image interpretation. As important physical components that influence the image quality of the display, luminance, contrast, resolution, and granularity are generally included. Among them, the resolution is especially an important component, which is greatly related to image quality.

FIGURE 7 — Outline of signal transfer for the horizontal direction of a CRT and an LCD. An LCD is different from the horizontal of a CRT in that the input signal is discrete and the LPFs does not exist. The horizontal direction of a CRT can be treated as well as the pure analog imaging devices.

In-field assessment of display resolution and noise: Performance evaluation of a commercialmeasurement system

Ehsan Samei
Esi Cleland
Hans Roehrig

Duke University, U.S.A.

Abstract — Two key metrics of image quality for high-fidelity displays, including medical displays, are resolution and noise. Until now, these properties have been primarily measured in laboratory settings. For the first time, a system consisting of a CCD camera and analysis software has been made commercially available for measuring the resolution and noise of medical displays in a clinical setting. This study aimed at evaluating this new product in terms of accuracy and precision. In particular, the project involved the measurement of the modulation transfer function (MTF) and the signal-to-noise ratio (SNR) of two medical imaging displays, one cathode-ray-tube (CRT) display and one liquid-crystal display (LCD) using this camera/software system. To assess the system's precision, measurements were made multiple times at the same setting. To check for accuracy, the results were compared with published values of the MTF and noise for the same displays. The performance of the system was also ascertained as a function of the focus setting of the camera. The results indicated that for the LCD, when the camera is focused within ±0.6 mm of the optimum focus setting, the MTF values lie within approximately 14% of the best focus MTF at the Nyquist frequency and 11% of the optimum overall sharpness (∫ MTF2 df).

Prior efforts have focused on methods for evaluating the image quality of medical displays in terms of display resolution and display noise. The standard guidelines for performance evaluation of electronic display devices, provided by the American Association of Physicists in Medicine, underscores the need for an easy, yet efficient way to quantitatively characterize these physical properties. However, until now, quantitative evaluation of display-resolution and noise-quality attributes has been largely limited to laboratory settings.

TABLE 1 — Characteristics of the display systems evaluated in the study.


In-field evaluation of the modulation transfer function and the signal-to-noise ratio of electronic-display devices

Hans Roehrig
Jerry Gaskill
Jiahua Fan
Chadwick Martin
John Greivenkamp
Ehsan Samei

University of Arizona, U.S.A.

Abstract — A charged-coupled device (CCD) camera, which was developed for in-field evaluation of the image quality of electronic-display devices [such as cathode-ray tubes (CRTs) and liquid-crystal displays (LCDs)] used for medical applications is described. Contrary to traditional cameras for display-image-quality evaluation, this CCD camera does not require a sophisticated x- y- z translation stage for mounting and adjustment. Instead, it is handheld and pressed by gentle pressure against the display screen. It is controlled by a software package which was originally developed for display calibration according to the DICOM 14 gray-scale standard display function (GSDF) This software package controls the camera gain when measurements are made at different display luminance, display test patterns, performs image analysis, and displays the results of the measurements and calculations. The work concentrated on the measurement of the modulation transfer function (MTF) and of the signal-to-noise ratio (SNR) per display pixel.

Electronic medical-display devices require routine monitoring of their image quality. Spatial resolution of CRTs and luminance of CRTs and LCDs vary systematically due to physical principles and by accident as a result of misadjustment. Display-evaluation guidelines such as those suggested by the AAPM Task Group 18 recommend mainly visual tests for assessment of the spatial resolution and noise property of electronic-display devices in the clinic. Quantitative evaluation of display spatial resolution and noise property using a high-performance CCD camera has been successfully done in the laboratory, but mostly off-line.


FIGURE 2 — The camera with a long focusing tube to accommodate use with an LCD.

Noise estimation and reduction on five medical liquid-crystal displays

Jiahua Fan
Hans Roehrig
Malur K. Sundareshan
Elizabeth A. Krupinski

University of Arizona, U.S.A.

Abstract — Liquid-crystal displays (LCDs) are replacing cathode-ray-tube (CRT) displays as primary diagnostic viewing devices in clinics. They exhibit higher spatial noise than CRTs, which can interfere with diagnosis and reduce the efficiency especially when subtle abnormalities are presented. A study by the authors on LCD spatial noise has recently been reported. A high-quality CCD camera was used to acquire images from the LCD. Noise properties were estimated from the digital-camera images. Then, an error-diffusion-based operation was applied to reduce the display spatial noise. The noise estimation and reduction results on five different medical-grade LCDs using the same study protocol is presented. These five different LCDs vary in terms of matrix size, pixel size, pixel structure, and vendors. The purpose of this work is to demonstrate that the LCD spatial-noise estimation and reduction scheme proposed earlier by the authors is valid, robust, and necessary for various medical-grade LCDs used in clinics today.

For noise property acquisition, a high-quality CCD camera was used to focus on the LCD and capture images. Noise properties were then estimated from the camera images using signal modeling. After noise estimation, an error-diffusion-based operation was applied to reduce the display spatial noise. The noise estimation and reduction results on five different medical-grade LCDs using the same study protocol are presented.


FIGURE 2 — Two LCD pixel types used in this study: left, a chevron-shaped LCD pixel; right, a cross-type LCD pixel.

A photographic technique for assessing the viewing-angle performance of liquid-crystal displays

Kenneth A. Fetterly
Ehsan Samei

Duke University, U.S.A.

Abstract — Liquid-crystal displays (LCDs) have notable variation in luminance and perceived contrast as a function of the angle from which they are viewed. Although this is an important performance issue for LCDs, most evaluation techniques for assessing this variation have been limited to laboratory settings. This study demonstrates the use of a photographic technique for such an evaluation. The technique is based on an actively cooled charge-coupled-device (CCD) detector in combination with a macro lens covering a circular angular range (θ) of ±42.5°. The camera was used to evaluate the luminance and perceived contrast properties of an LCD. Uniform field images corresponding to 17 equally spaced gray-scale values in the digital driving level (DDL) range of the display system were acquired. The 12-bit gray-scale digital images produced by the camera were converted to luminance units (cd/m2via the measured luminance vs. DDL response function of the camera.

The luminance and contrast properties of an LCD were measured as a function of viewing angle using a CCD detector and a 4-mm lens. The radially symmetric ±42.5° field of view of the lens is sufficiently wide to characterize viewing-angle dependence of electronic displays used for applications that require high-quality image presentation. The inherent 2-D acquisition properties of the camera provide simultaneous measurement of luminance, and thereby contrast, over a wide range of viewing angles.

FIGURE 5 — Luminance (cd/m2) map for DDL = 127. The contour lines are separated by 1 cd/m2.

Preferred balance between luminance and color gamut in mobile displays

Jun Xia
Han Chun Yin
Ingrid Heynderickx

Southeast University, China

Abstract — To improve the image quality of a mobile display, the balance between color-gamut size and luminance was studied in two subjective experiments. The first experiment was performed during the Asian Society for Information Display (ASID) conference in Nanjing, February 2004. Nearly 600 participants ranked the quality of images displayed for fixed combinations of color-gamut size and display luminance on small color supertwisted nematic (CSTN) and thin-film transistor (TFT) twisted-nematic (TN) displays. In the second experiment, a broader range of color-gamut sizes and luminance levels were simulated on a cathode-ray-tube (CRT) display, and 20 participants were asked to score perceived image quality. The results of these experiments were used to model image quality as a function of color-gamut size and display luminance for images differing in the level of chromaticity of their content. This model can be used to estimate the increase in luminance required to compensate for a reduction in color-gamut size.

There is a simple solution to the limited color-gamut size of a display, which is narrowing the spectral bandwidth of the color filters of the LC cell. As a consequence, only light from a limited range of wavelengths is transmitted with the result that colors become more saturated. However, this is at the expense of the display's brightness, which was also shown to be an important attribute of the overall image quality. One option is to increase the display's brightness by increasing the lumen output of the display's backlight. However, this negatively affects power consumption.


FIGURE 1 — Image material used in the experiments: (a) Fruit, (b) Cloth, (c) Portrait, and (d) Sand.

Quality assessment of false-colored fused displays

Timothy D. Dixon
Eduardo Fernández Canga
Stavri G. Nikolov
Tom Troscianko
Jan M. Noyes
Dave R. Bull
C. Nishan Canagarajah

University of Bristol, U.K.

Abstract — The problem of assessing the quality of fused images (composites created from inputs of differing modalities, such as infrared and visible light radiation) is an important and growing area of research. Recent work has shown that the process of assessing fused images should not rely entirely on subjective quality methods, with objective tasks and computational metrics having important contributions to the assessment procedure. The current paper extends previous findings, applying a psychophysical selection task, metric evaluation, and subjective quality judgments to a range of fused surveillance images. Fusion schemes included the contrast pyramid and shift invariant discrete wavelet transform, the complex wavelet transform, and two false-coloring methods. In addition, JPEG2000 compression was applied at two levels, as well as an uncompressed control. Reaction time results showed the contrast pyramid to lead to slowest performance in the objective task, whilst the presence of color greatly reduced reaction times. These results differed from both the subjective and metric results. The findings support the view that subjective quality ratings should be used with caution, especially if not accompanied by some task.

The process of combining two or more images of differing modalities (as shown in Fig. 1) is designed to bring about two goals: an increase in the total amount of useful information presented through combining complementary elements of the inputs and a reduction of redundant information for a given task through processes such as de-noising, and not introducing new artifacts. Thus, one might look to combine CT and MR medical scans to improve clinical detection rates, combine high-spatial-resolution monochrome remote-sensing images with low-spatial multispectral images to aid land-cover classification, or combine infrared and visible-light surveillance imagery to aid situational awareness and target detection.


FIGURE 1 — Representation of a generic fusion process. Visible-light image is combined with infrared image using the discrete wavelet transform to create the fused image.

Black-level offset: Characterization and correction

Jacobus Besuijen

University of Twente, The Netherlands

Abstract — The correct setting of the black level is an important step in the (re)calibration of an electronic display. This study looks at the consequences of black-level offset, the possibilities for display characterization with offset, offset correction, and the ability of average untrained users to visually correct the black-level setting with the contrast and brightness controls on the display. In an experiment, 32 subjects were asked to optimally set the black level according to two types of instructions (short and extensive, between subjects) under two levels of illumination (low and office, between subjects) for two types of displays (CRTs and LCDs, within subjects). Most subjects were not able to set the black level near optimal for either display, with any combination of instruction and illumination level. The LCD did not have an optimal black level. For the CRT, optimal black level did not provide minimal differences with the sRGB standard tone reproduction curve. For color-management purposes, there are two problems related to the black-level offset. If the display has to be used in an illuminated environment where veiling glare cannot be avoided, a black-level offset might be desired to preserve the color differences within the darker parts of the image. In that case, a correct characterization is needed to provide an optimal color fidelity for the circumstances. If the level of veiling glare is low, then the black-level offset should be corrected. A correct characterization of the display can be helpful in determining the brightness and contrast control settings that produce the lowest black-level offset.


FIGURE 3 — Differences in CIE lightness from sRGB for the LCD in the experiment. BW gives the overall difference for neutral stimuli. Red, green, and blue give the differences for the separate primaries corrected for black level, which shows the difference if only black-level offset were present.

Calibration of diagnostic monitors: Theoretical determination of optimal luminance settings

Magnus Båth
Patrik Sund
Linda Ungsten
Lars Gunnar Månsson

Sahlgrenska University Hospital,Sweden

Abstract — Common practice today is to calibrate diagnostic monitors according to the gray-scale standard display function (GSDF) described in DICOM part 14. However, the GSDF is based on the assumption of variable adaptation of the human-visual system (HVS). It is well known that the HVS adapts to the average quantity of light falling on the retina, so-called fixed adaptation. For the luminance setting of a monitor, the effect of fixed adaptation is of interest. The wider the luminance range of the monitor, the larger the number of available just-noticeable differences (JNDs). However, at the same time, the sensitivity of the HVS to the average contrast change is decreased since it occurs at a luminance level further away from the adaptation luminance. A computer program was therefore written which takes the effect of the fixed adaptation into account by determining the number of effective JNDs for a given luminance setting of a monitor. The probability of detecting each change in presen-tation value is then calculated from the distribution of effective JNDs.

A display system was characterized by its maximum and minimum luminance settings (Lmax and Lmin, respectively, not to be mistaken for the maximum and minimum possible luminances) and the number of available digital driving levels (DDLs). A computer program was written which initially distributes the number of available JNDs equally over the DDLs so that each unit DDL change results in a luminance change corresponding to a constant number of JNDs. This corresponds to the assumption that the monitor can be perfectly calibrated internally, which requires that there are enough luminance levels available for each DDL to result in the exact desired luminance level for any setting of Lmax and Lmin.


FIGURE 2 — Pave as a function of Lmin (x axis) and Lmax (y axis) for an adaptation luminance of (c) 100 cd/m2 and (d) 1000 cd/m2. The calculations are based on an 8-bit system with the standard DICOM test pattern. Both Lmin and Lmax range from 1 to 1000 cd/m2on logarithmic axes.

Generation of low-contrast sinusoidal test patterns on a high-brightness display

Patrik Sund
Magnus Båth
Linda Ungsten
Lars Gunnar Månsson

Sahlgrenska University Hospital, Sweden

Abstract — By using a technique of addressing individual subpixels, low-contrast sinusoidal test patterns were generated on a high-brightness color LCD. Detection tasks were performed using 2AFC experiments, and the value obtained for the contrast threshold was found to agree well with previously published data. As long as the shape of the pattern was close to sinusoidal, a good correlation was found between pattern contrast and detection rate. The influence of the eye-adaptation luminance level was studied and compared to the f-factor published by Barten. Luminance non-uniformities in the background did not cause any significant change in the detection threshold.

Human contrast detection is complex and much is still unknown when it comes to real clinical situations where the task is far from detecting a well-known test pattern on a homogenous background. Several studies under realistic conditions are needed in which not only the test pattern size and shape are varied, but also background properties such as luminance range and spatial distribution. Because most clinical sites nowadays use LCDs for softcopy reporting, it is also necessary to study the influence of viewing angle, display bit-depth, and luminance stability.


FIGURE 2 — Test patterns used for the two studies. The background average luminance was the same for both patterns (20 cd/m2). A high-contrast reference pattern was always visible, either at the bottom of the screen (homogeneous) or to the right of the screen (striped). The reference patterns are not included in this picture.

An alternate method for using a visual discrimination model (VDM) to optimize softcopy display image quality

Dev P. Chakraborty

University of Pittsburgh, U.S.A.

Abstract — Researchers have developed visual discrimination models (VDMs) that can predict a human observer's ability to detect a target object superposed on an image. These models incorporate sophisticated knowledge of the properties of the human-visual system. In the predictive approach, termed conventional VDM usage, two input images with and without a target are analyzed by an algorithm that calculates a just-noticeable-difference (JND) index, which is a taken as a measure of the detectability of the target. A new method of using the VDM is described, termed channelized VDM, which involves finding the linear combination of the VDM-generated channels (which are not used in conventional VDM analysis) that has optimal classification ability between normal and abnormal images. The classification ability can be measured using receiver operating characteristic (ROC) or two alternative forced choice (2AFC) experiments, and in special cases they can also be predicted by signal detection theory (SDT) based model-observer methods.


FIGURE 1 — This figure shows the different background regions studied in this work. Shown are (a) a nodule-free mammographic region, (b) an uncorrelated Gaussian noise region, and three simulated power-law noise regions with different degrees of spatial correlation described by the β parameter (see text) (c) β =1, (d) β = 2, and (e) β = 3. Note the similarity of the mammogram-region (a) and the simulated region (e). The same seed was used to generate all simulated images in this example.

Using the human observer to assess medical image display quality

Elizabeth A. Krupinski

University of Arizona, U.S.A.

Abstract — Image quality is an important component in the evaluation of medical imaging systems and can be measured at a number of points between acquisition and display. However, from the perspective of the clinician, the quality of the image that is presented to the eye–brain system is the most critical since this is what the diagnostic interpretation is derived from. Some ways to assess the design and optimization of display systems in reference to the human observer will be discussed. Various approaches to study this issue, including Receiver Operating Characteristic studies, modeling of the human-visual system, and the use of eye-position tracking methods, are included. The goal is to demonstrate not only the importance of assessing the impact of display image quality on diagnostic performance, but also on how it can affect workflow and user-comfort levels.

The problem in medical imaging is that the definition and hence the quantitative and objective measurement of image quality differs depending on how the question is posed and to whom. The imaging chain is a complex sequence of events during which the image is susceptible to a wide array of potentially degrading factors. The chain actually begins with the object being imaged (the patient) and pro-ceeds through acquisition, transmission, display, and finally processing by the human eye–brain system. Since the images may need to be reviewed in the future, storage and retrieval also become important steps in the chain.


FIGURE 3 — Model vs. human performance in the on- and off-axis viewing conditions for LCD and CRT displays.

Guiding principles for high-quality moving pictures in LCD TVs

T. Yamamoto
S. Sasaki
Y. Igarashi
Y. Tanaka

Hitachi, Ltd., Japan

Abstract — The moving-picture quality of several LCD modules was evaluated by using the quantitative parameter, normalized blurred edge width (N-BEW), or the N-BEW value normalized by time (N-BET), measured and calculated by the developed time-sequence-image integration system which has taken LCD-response characteristics and human-vision characteristics into consideration. The quality of several LCD TVs is also discussed by using subjective evaluation and the unified quantitative parameter moving-picture response time (MPRT), which is based on N-BEW. According to the experimental and calculated results, it is clear that the value of N-BET can express moving-picture quality, which depends on the liquid-crystal response time and the hold-type character of LCDs. Also, it is confirmed that the value of MPRT can express the moving-picture quality by comparison with subjective evaluation. The target values of MPRT and N-BET for the motion-blur-less picture are deduced by extrapolating the subjective evaluation results. Then, guidelines to improve the moving picture quality are demonstrated.

A perceived blurred motion image on an LCD is not the same as a moving-picture photograph because of the human-vision characteristics and hold-type display affect. If the response time of an LC is ideally fast or zero, displayed light on LCDs is sustained in a frame period of about 1/60 sec. Thus, LCDs are hold-type displays. Moving images are perceived as blurred by an inconsistency between the sustained light of hold-type displays and eye movement. The mechanism for the perceived motion blur is as follows. Two properties of human vision are assumed when watching a moving image: (1) The viewpoint accurately tracks the viewed object by eye movements. (2) Light stimulus within 1/60 sec is completely integrated into the visual system. It has been reported that both assumptions are usually valid under certain conditions.


FIGURE 1 — Temporal sequence of displayed image on a hold-type display as related to the human-vision process. (a) Response time of 0 msec. (b) Response time of about one frame period.

HD motion-picture evaluation method for overdrive technology in the frequency domain

Haruhiko Okumura
Masahiro Baba
Kazuki Taira
Makiko Okumura

Toshiba Corp., Japan

Abstract — The development of an advanced level-adaptive overdrive (A-LAO) method applicable to a full-HD LC projector with 1.84 Mpixels, which reduced the gray-level response time to less than 16 msec, is introduced. In addition, it is shown that a response of less than 8 msec can be achieved by combining the A-LAO method with a frame interpolation method (120-Hz refresh). A new motion-picture evaluation method using frequency-domain analysis, in other words, perceived bandwidth instead of the conventional time-domain-analysis response time evaluation, is reported.

In recent years, many driving concepts similar to our original LAO method and recursive LAO method have been developed, and they indicate the usefulness of our original concept. However, those methods have been applied to LCDs with standard definition. To realize high-definition LC TV with using overdrive technology, a high-speed low memory capacity and low-noise overdrive technology was necessary to improve the artifacts caused by higher bandwidth. We have developed an advanced LAO (A-LAO) method for full-HD LCD TV with half memory size by using a novel image-compression method without noise degradation by the motion detective technique for the first time in 2002.

FIGURE 2 — Advanced level adaptive overdrive (A-LAO) circuit block diagram.

Dynamic gamma: Application to LCD motion-blur reduction

Xiao-fan Feng
Hao Pan
Scott Daly

Sharp Laboratories of America, U.S.A.

Abstract — A metric, "dynamic gamma," to quantitatively evaluate the dynamic temporal response of an LCD device is proposed. Dynamic gamma, associated with 2-D plots, is more suitable for quantitatively characterizing the dynamic characteristics of an LC panel. The dynamic gamma metric was applied to improve the temporal response of LCDs. From dynamic gamma data, overdrive tables can be derived. Dynamic gamma can also be used to evaluate the effectiveness of overdrive. With a second-order dynamic gamma, the performance of different overdrive algorithms can be quantitatively assessed. The dynamic gamma metric was also applied to backlight flashing and developed a time adaptive overdrive algorithm. The new algorithm reduces the ghosting artifact due to the timing mismatch between LCD driving and backlight flashing. Experimental results from a simulated tracking camera confirms the advantages of the new algorithm designed using dynamic gamma.


FIGURE 13 — Measured temporal response of a transition (white to black and then to gray).