Biomed Imaging Interv J 2006; 2(3):e38
doi: 10.2349/biij.2.3.e38
© 2006 Biomedical Imaging and
Intervention Journal
Technical Note
Development of a randomised contrast detail digital phantom for observer detectability study
MS Nizam, BSc, MMedPhys,
KH Ng, PhD, DABMP,
BJJ Abdullah, MBBS, FRCR
Department of Biomedical Imaging, Faculty of Medicine, University of Malaya, Kuala Lumpur, Malaysia
ABSTRACT
The accuracy and the efficacy of radiological diagnosis depend, to a large
extent, on the conditions under which radiographs and images
are viewed. This mainly involves the luminance of the display
devices and the ambient room illumination. We report a perceptual
study to investigate the relationship between detectability
and monitor luminance as well as ambient illuminance. A statistical
test pattern was used in this study, and the test pattern was
developed using Microsoft® Visual Basic 6. The test pattern
contained a set of randomised contrast detail objects, that
is, disks of different diameters (0.7, 1.0, 1.4, and 2.0 mm)
and contrasts against a black background (2.7, 3.9, 5.5, and
7.8%), simulating lesions in digital images. The receiver operating
characteristic (ROC) analysis was used in this study. The results
indicated that a set of optimal viewing conditions exists and
that it has a significant effect on detectability performance.
© 2006 Biomedical Imaging and Intervention Journal. All rights
reserved.
Keywords: Image quality, contrast detail, perception,
receiver operating characteristics
INTRODUCTION
The accuracy and the efficacy of radiological diagnosis
greatly depend on the conditions under which radiographs and images are viewed.
This mainly involves the luminance of the display devices and the ambient room
illumination [1-5]. A number of computer programs have also been used to
evaluate image quality on softcopy display [6-8]. In this study, a randomised
contrast detail digital phantom was developed to study the effect of
CRT (cathode ray tube) display luminance and ambient illuminance on the
perception of the observer. The digital phantom was so designed that the
location of the pathology simulators within the phantom changed for every new
use to ensure objectivity.
The digital phantom is available for download at http://www.biij.org/2006/3/e38/e38.exe.
MATERIALS AND METHODS
The digital phantom was developed using Microsoft® Visual
Basic 6 to generate 16 pathology simulators in the form of disks
of different diameters (0.7, 1.0, 1.4 and 2.0 mm) and contrasts
against a black background (2.7, 3.9, 5.5 and 7.8%) on CRT display
monitor. These values were calculated with multiplicative factor
of  according to Cd =
constant, where C is the contrast of the test element
with the background and d is the diameter of test element
[9,10].
The digital phantom consisted of 6x6 squares (a total of 36
squares) at the centre of a 1280x1024 resolution CRT display monitor. The
squares were labeled according to column and row from 1 to 6. Lines of the
squares were 20% brighter than the black background. The 16 disks were
generated randomly, and one disk could occupy only one square (Figure 1). A
text document file with information about disk location, size, and contrast was
also generated in the computer’s hard disk drive (C:\) each time the software was run (Figure 2).
Ten observers, who were radiographers with minimum three
years of working experience, were asked to detect the disks from images created by
the software. They were asked to use a 5-point scale, from 1 – definitely
absent to 5 – definitely present. Their answers were compared with the correct
answers provided by the text document file.
Display luminance measurements were taken according to the
method recommended by Parsons et al. [11] using a luminance level meter (Mavo-Monitor,
Gossen-Metrawatt Gmbh, Nürnberg, Germany). The contrast level was set to 100%.
After 30 minutes of warm up, the 100% grey-level square of a SMPTE test pattern
was zoomed so that it filled the display area. The screen was divided into nine
squares and measurements of the luminance output were taken at the centre of
each of these squares. The luminance was adjusted to the required level using
the brightness control.
Ambient illuminance measurements were taken according to the
method recommended by Moores et al. [12] using a photometer (model PMLX, Quantum
Instruments Inc., New York, USA). All types of display were switched off, and
the photometer was used to measure the level of room illumination at a point 30 cm from
the display. The room illuminance level was adjusted using a light dimmer.
The phantom was tested on a 21-inch colour CRT monitor (Sony
model GDM-500PS, Sony Electronics Inc., California, USA) with 1280x1024
resolution and 32 bit at 85 Hz refresh rate. The CRT monitor luminance was 120.0±0.9 and 80.0±0.9
cd/m2, respectively, each viewed under 30.0±0.1, 50.0±0.1,
and 70.0±0.1 lux room illumination. Observers
were given approximately 10 minutes for their eyes to adapt to the room
illumination before commencing to evaluate each test image.
RESULTS
Overall ROC curves and area under the curve values (Az)
were perfect or almost perfect for big size disks (2.0 and 1.4 mm), but as the
disk size decreased observer performance also decreased. Higher ambient
illumination decreased observer performance, and its effect was more pronounced
with lower display luminance (80 cd/m2) compared with higher display
luminance (120 cd/m2). Figures 3 and 4 show the ROC curves and Az
values obtained for 1.0 mm disks on 80 and 120 cd/m2 CRT
monitors, respectively.
DISCUSSION AND CONCLUSION
The result of this study concerning the effect of display luminance
and ambient illuminance is in agreement with previous findings
[1-5]. Based on the result, a high display
luminance with minimum reasonable ambient illuminance is recommended
to optimise softcopy reporting. Other relevant factors that
may influence the perception tasks involved in the radiology
reading room session have been listed by Wang and Langer [13].
They are: (1) spatial and contrast resolution of the display
device; (2) brightness and the displayed luminance range of
the monitor (or film/viewbox); (3) uniformity of the display
system luminance; (4) extraneous light in the reading room (such
as bright, unmasked areas on the monitor and light reflected
off the monitors); (5) displayed field size (field of view);
(6) viewed object orientation; (7) image motion and flickering
of display device; (8) signal to noise ratio of the displayed
image; (9) magnification and zooming functions; and (10) user
interface of the workstation.
The performance of softcopy reporting has become an
important issue due to the rapid introduction of digital radiology in hospitals.
This randomised contrast detail digital phantom is another valuable tool for
observer detectability study and quality control of softcopy display. It could
be used to evaluate the new generation of flat screen displays being used
increasingly in radiology departments.
ACKNOWLEDGEMENT
The authors would like to thank all the radiographers participating in this study as observers.
REFERENCES
-
Roehrig H, Krupinski E. Image Quality of CRT displays and the effect of brightness of diagnosis of mammograms. J Digit Imaging 1998;11(3 Suppl 1):187-8.
[Medline]
-
Flynn MJ, Badano A. Image quality degradation by light scattering in display devices. J Digit Imaging 1999;12(2):50-9.
[Medline]
-
Hemminger BM, Dillon AW, Johnston RE, et al. Effect of display luminance on the feature detection rates of masses in mammograms. Med Phys 1999;26(11):2266-72.
[Medline]
-
Roehrig H, Krupinski EA, Furukawa T. Evaluation of a flat CRT monitor for use in radiology. J Digit Imaging 2001;14(3):142-8.
[Medline]
-
Chakrabarti K, Kaczmarek RV, Thomas JA, et al. Effect of room illuminance on monitor black level luminance and monitor calibration. J Digit Imaging 2003;16(4):350-5.
[Medline]
[CrossRef]
-
Wang J, Peng Q. An interactive method of assessing the characteristics of softcopy display using observer performance tests. J Digit Imaging 2002;15 Suppl 1:216-8.
[Medline]
[CrossRef]
-
Davydenko G, Gurvich V, Smekhov M. Application of StatPhantom software for image quality evaluation. J Digit Imaging 2002;15 Suppl 1:219-20.
[Medline]
[CrossRef]
-
Norrman E, Gardestig M, Persliden J, et al. A clinical evaluation of the image quality computer program, CoCIQ. J Digit Imaging 2005;18(2):138-44.
[Medline]
[CrossRef]
-
Rose A. Vision: Human and Electronic. New York: Plenum Press, 1973.
-
Burgess AE. The Rose model, revisited. J Opt Soc Am A Opt Image Sci Vis 1999;16(3):633-46.
[Medline]
-
Parsons DM, Kim Y, Haynor DR. Quality control of cathode-ray tube monitors for medical imaging using a simple photometer. J Digit Imaging 1995;8(1):10-20.
[Medline]
-
Moores BM, Watkinson SA, Henshaw ET, Pearcy BJ. Practical Guide to Quality Assurance in Medical Imaging. Chichester: John Wiley & Sons, 1991.
-
Wang J, Langer S. A brief review of human perception factors in digital displays for picture archiving and communications systems. J Digit Imaging 1997;10(4):158-68.
[Medline]
Received 24 November 2005; received in revised form 25 June 2006; accepted 19 July 2006
Correspondence: Department of Biomedical Imaging,
Faculty of Medicine, University of Malaya, 50603 Kuala Lumpur,
Malaysia. Tel.: +603-7949 2069; Fax.: +603-7958 1973;
E-mail: shahrun@um.edu.my
(Muhammad Shahrun Nizam).
Please cite as: Nizam MS, Ng KH, Abdullah
BJJ,
Development of a randomised contrast detail digital phantom for observer detectability study, Biomed Imaging Interv J 2006; 2(3):e38
<URL: http://www.biij.org/2006/3/e38/>
This article has been viewed 1653 times.
|
|
 |
|
Leave a comment