| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
1 All authors: Department of Diagnostic Radiology, Asan Medical Center, University of Ulsan College of Medicine, 388-1 Poongnap Dong Songpa Ku, Seoul 138-736, Korea.
Received October 26, 2000;
accepted after revision February 22, 2001.
Address correspondence to K. S. Cho.
Abstract
 Top Abstract IntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
MATERIALS AND METHODS. We compared observer performance for detecting urinary calculi using three sets of radiographs—hard-copy images, soft-copy images displayed on a LCD monitor (1280 x 1024 bits), and soft-copy images displayed on a high-resolution video monitor using receiver operating characteristic curve analysis with a continuous rating scale. Computed radiography was archived with a 2140 x 1760 pixel resolution and a 10-bit depth. The selected data set included 62 images: 27 images showing proven urinary calculi smaller than 6 mm and three in number, and 35 images containing no proven abnormalities. Eleven radiologists (three genitourinary radiologists and eight general radiologists) participated in the study. Interpretations of three sets of randomly distributed radiographs were performed individually in three separate sessions at 1-week intervals.
RESULTS. No statistically significant differences were found in the area under the receiver operating characteristic curve for detecting urinary calculi or in the interpreting times between soft-copy and hard-copy images; the mean areas under the receiver operating characteristic curve of hard-copy images, soft-copy images displayed on an LCD monitor, and soft-copy images displayed on a high-resolution video monitor were 0.579, 0.610, and 0.732, respectively. However, soft-copy images showed relatively improved diagnostic accuracy among less experienced radiologists (p < 0.05).
CONCLUSION. For detecting urinary calculi, soft-copy images offered a diagnostic accuracy similar to or slightly more accurate than that of hard-copy images obtained in a laser-printed film-based environment. The diagnostic performance with soft-copy images viewed on an LCD monitor was comparable to that of soft-copy images viewed on a high-resolution video monitor.
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
A recent report showed that large differences in luminance exist between conventional viewbox displays and most soft-copy displays [7]. Because of a lack of cost-effective soft-copy viewing technology that yields brighter images while preserving resolution and contrast sensitivity, much effort has been expended in an attempt to define nonuniformities in luminance. If PACS is to succeed in the clinical environment, both the luminance factor and the resolution factor must be taken into account. Until now, high-resolution video monitors of 2K x 2K have been widely used as soft-copy viewing systems for PACS. However, in the environment of filmless imaging, the universal use of this expensive monitor in clinical practice may not be cost-effective. Therefore, we must consider an alternative to the high-resolution video monitor that is cost-effective and that can present images as effectively as the high-resolution monitor.
The purpose of this study is to compare observer performance for the detection of urinary calculi using computed radiographs with different display formats and different viewing systems.
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
The digital data were also sent to a PACS server (Radpia; HyundaeTech, Seoul, Korea) and distributed to display workstations. All images were down-loaded onto the local hard disk drives of the display workstations before being viewed by a radiologist. We used two viewer systems for soft-copy display: a high-resolution video monitor with 2000 x 2500 x 8 bit pixels (BACCO; Dataray, Denver, CO) used in a darkened room for side-by-side image display, and a liquid-crystal-display (LCD) monitor with 1280 x 1024 x 8 bits pixels (870LE; LG, Seoul, Korea). The monitors operated at 71 Hz in an interlaced mode and had a maximum brightness level of 100 foot-lamberts. The gray scale of digital images was modified by means of 10- to 8-bit (storage phosphor image) look-up table. To eliminate the differences between two of the same monitors, their maximal brightness was adjusted to be the same. The soft-copy images were displayed without unsharp masking. Only window width and image level were automatically optimized by the customized program. The observers were allowed to adjust the brightness and contrast of the images. Localized magnification of each image was also permitted. For this observer performance study, patient identification was obscured on all images and was replaced by a sequence number.
Image Collection
During a period of 3 months, abdominal radiographs of patients who were
diagnosed with either renal or ureteral stone disease were obtained by
searching the medical records of our institute. The diagnosis of urinary stone
disease was based on the following criteria the presence of a stone directly
confirmed by surgery or lithotripsy; the presence of a radiopaque stone
detected on plain radiography and on excretory urography but not seen on
follow-up conventional radiography or on excretory urography after painful
urination or extracorporeal shock-wave lithotripsy; and the presence of a
radiopaque stone confirmed on retrograde pyelography or excretory urography
plus unenhanced CT. Patients who were suspected of having urinary calculi but
who had been proven not to have calculi were then selected for the control
group by searching both the radiologic and medical records. These patients
showed no abnormal findings on excretory urography plus sonography or CT. Each
image was screened by the study clinical coordinator, a board-certified
radiologist with at least 7 years of clinical experience. Screening was done
to ensure that selected images met the following specifications: the image was
of acceptable diagnostic quality; subtle findings and normal findings were
shown; and the selected images were not correlated with other images already
collected for the evaluation.
The initial screening resulted in the selection of 114 images. Approximately half of these were selected using the digital soft-copy display as the viewing technique; the remaining half were selected using the digital hard-copy film during the screening. All patients underwent excretory urography (n = 85), sonography (n = 107), or CT (n = 79).
These images were screened a second time by a verification committee composed of two experienced urologic radiologists who had access to all information related to each patient, including the clinical history, historical data, and images obtained using other radiologic techniques. Thirty-one images were eliminated because of the presence of too-obvious calculi on the radiograph, such as a stone larger than 6 mm in its long axis or clustered calculi numbering more than three. Twenty-one images were also excluded by the committee because the members could not reach a consensus decision.
Finally, after the two screenings, 62 images remained, consisting of the disease group (n = 27) and the control group (n = 35). These images were assigned study identification numbers, and the original hard-copy films were collected and organized. The digital data were moved to an independent folder in the PACS hardware and loaded onto a high-speed parallel transfer disk (CD-ROM).
Study Performance
Eleven radiologists compared these three sets of images using a
standardized protocol. Three of the 11 observers were senior urologic
radiologists, each with 20 years' experience, and two had fellowships in
urologic imaging. Their clinical practices involved urologic imaging almost
exclusively. The other eight observers were senior residents. All observers
were accustomed to PACS because they used it in their daily practice.
Interpretations of three sets of randomly distributed radiographs were
performed individually in three separate sessions held at 1-week intervals.
Each observer was unaware of the patient's history and interpreted each image
independently.
The workstation was set up in an isolated room with constantly dimmed room lighting. The interpretation conditions closely simulated the clinical interpretation environment. For the soft-copy display, all images were loaded onto the hardware of the PACS workstation, on which it took less than 1 sec for the observer to display any image. The software package in the workstation included such features as advance to the next case, real-time contrast and brightness adjustment, zoom via pixel replication, and contrast reversal. All digital operations were performed with a mouse.
The presence or absence of urinary calculi was evaluated using the following scoring system: 1, definitely negative; 2, probably negative; 3, indeterminate; 4, probably positive; and 5, definitely positive. Second, the image quality for interpretation, which focused on anatomic conspicuity, was evaluated using the following scoring system: 1, poor; 2, fair; and 3, excellent for diagnosing. Finally, the required time for interpretation in each session was recorded individually. These responses were recorded and resorted by each system for statistical analysis.
Statistical Analysis
To evaluate observer preference in detecting urinary calculi and taking
into account the various viewing methods, the CORROC2 program (University of
Chicago, Chicago, IL) was used to calculate the receiver operating
characteristic curve area and its standard deviation for a given observer's
results. The average receiver operating characteristic areas from all
observers using various visualization formats wee compared using the paired
Student's t test. The difference in image quality for each observer,
according to the different viewing methods, was also compared using the paired
t test (p < 0.05).
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
|
|
|
|
|
|
|
In evaluating image quality, no statistically significant difference was seen among the three imaging display methods (Table 2). However, less experienced radiologists seemed to prefer the soft-copy images on the high-resolution video monitor compared with hard-copy film or soft-copy images viewed on the LCD monitor.
|
The mean times spent per image for each observer are provided in Table 3. The differences in the mean interpretation times for the different image types were within the range of the intraobserver variability. Only two radiologists performed more slowly with the soft-copy display than with hard-copy film.
|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
In this study, we compared observer performance for detecting urinary calculi on computed radiographs with digital hard-copy film versus an interactive soft-copy display. Our results agree with those in similar experiments on chest imaging comparing 2K x 2K soft-copy versus digital hard-copy computed radiographic images [1]. Although no statistically significant difference was seen between viewing images on digital hard-copy versus soft-copy, our observers performed better when using 2K x 2.5K soft-copy displays than when using digital hard-copy film for detecting urinary calculi. Interestingly, this improved performance occurred more often with less experienced radiologists than with more experienced radiologists. This difference may be a result of the extensive experience of our observers in viewing images from soft-copy displays, particularly the less experienced radiologists, who are more familiar with soft-copy display than with hard-copy film. The familiarity of these less experienced radiologists with the digital operation in soft-copy viewing seemed to lead them to prefer soft-copy display.
The interpretation of soft-copy images may be more time-consuming than the hard-copy viewing because of additional digital operations associated with viewing soft-copy images. In clinical practice, if soft-copy viewing is proven to be time-consuming because of digital operations, the field of viewing will be limited. However, despite all the expectations, no significant difference was seen in terms of time required for interpretation with respect to the different display systems in this study. Only two radiologists performed more slowly with soft-copy than with hard-copy images. One of the two was a less experienced radiologist and the other was an experienced one. This fact suggests that the required time for interpretation was decided not by experience but by familiarity with digital operations or the individual personality.
The observers' performances with soft-copy images on an LCD monitor (1K x 1K) were comparable to those with soft-copy images on a high-resolution video monitor (2K x 2.5K) despite the difference in resolution. These results suggest the feasibility of the LCD monitor as an alternative to the more expensive high-resolution monitor in the extensive medical environment using PACS. However, replacement of the entire high-resolution video monitor system by an LCD monitor in a radiology department should be considered carefully before making the change because of the inherent limitations of the difference in spatial resolution. The wide scale of the receiver operating characteristic study is necessary to determine the feasibility of the LCD monitor as a soft-copy display system in routine practice.
During the recent rapid development of electronic and computer technology, digital radiologic detectors have undergone considerable investigation and development [9,10,11,12]. With these advances in digital radiography, the use of PACS will continue to expand rapidly. In digital radiography, the pixel size of the detector and the resolution of the monitor are important factors for improving the spatial resolution of the image. Further investigation will be necessary to determine which flat-panel detector and which soft-copy display monitor are preferable in a reliable diagnostic environment.
In summary, in our study the soft-copy display of computed radiographs performed better than hard-copy images for detecting urinary calculi. In addition, the diagnostic performance with soft-copy images viewed on an LCD monitor was comparable to that of soft-copy images viewed on a high-resolution video monitor. These results suggest that computed radiographs viewed in a soft-copy PACS environment should result in diagnostic performance similar to or slightly more accurate than that obtained in a laser-printed film-based environment. Our results also suggest the distinct clinical feasibility of the LCD monitor for viewing soft-copy images.
Acknowledgments
We thank Bonnie Hami, Department of Radiology, University Hospitals Health
System, Cleveland, OH, for editorial assistance in preparing the
manuscript.
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
CiteULike 
Complore 
Connotea 
Del.icio.us 
Digg 
Reddit 
Technorati What's this?
This article has been cited by other articles:
|
|
 |
|
  J. E. Roos, B. Chilla, M. Zanetti, M. Schmid, P. Koch, C. W. A. Pfirrmann, and J. Hodler MRI of Meniscal Lesions: Soft-Copy (PACS) and Hard-Copy Evaluation Versus Reviewer Experience. Am. J. Roentgenol., March 1, 2006; 186(3): 786 - 790. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 C. D. Johnson Hard- versus Soft-Copy Interpretation Radiology, June 1, 2003; 227(3): 629 - 630. [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
