| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
1 Department of Radiology, University of Vienna, Währinger Gürtel
18-20, Vienna A-1090, Austria.
2 Department of Biomedical Engineering and Physics, University of Vienna, Vienna
A-1090, Austria.
3 Ludwig Boltzmann Institute for Clinical and Experimental Radiologic Research,
University of Vienna, Vienna A-1090, Austria.
Received July 8, 2002;
accepted after revision May 14, 2003.
Address correspondence to C. M. Schaefer-Prokop
(mathias.prokop{at}univie.ac.at).
Abstract
 Top Abstract IntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
MATERIALS AND METHODS. A total of 131 catheter fragments were superimposed over 10 bedside chest radiographs obtained with storage phosphor technology. Images were displayed on a clinical intensive care unit viewing station (color cathode-ray tube monitor, 21 inch [53 cm], 1280 x 1024 matrix) and were independently evaluated by five radiologists. The number of catheter fragments per image varied between 12 and 14, with an approximately equal distribution in high- and low-absorption areas. Detectability of catheter fragments was assessed under subdued and bright ambient light conditions with and without interactive adjustment of window width and level.
RESULTS. Under subdued light, the detection rate of catheter fragments was significantly higher than under bright light (51.8% vs 56.6%, p < 0.05). Interactive window setting adjustment significantly increased the detection rate from 52.5% to 60.8% (p < 0.05) under subdued light and from 47.9% to 55.6% (p < 0.05) under bright light. With adjustment of window settings, the difference between the detection rates under subdued light (60.8%) and under bright light (55.6%) did not reach statistical significance.
CONCLUSION. Detection of catheters on soft-copy display is significantly decreased by bright ambient light, an effect that can be largely compensated for by means of interactive adjustment of window settings.
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
As medical imaging moves strongly toward digital technology, more and more health care facilities are converting to digital filmless hospital and radiology information management. Within that scope, the soft-copy presentation of medical images gains increasing relevance for an efficient and cost-effective hospital organization.
High-resolution gray-scale cathode-ray tube display is the current gold standard for soft-copy display. Although most medical images are monochromatic, they can also be presented on color cathode-ray tube display systems. Color monitors have become the most common displays for Web-based or remote viewing by clinicians. These monitors usually have a lower spatial resolution than monochromatic cathode-ray tube monitors and are considerably less expensive [3].
Among other factors, spatial resolution, luminance, and contrast resolution of the monitor display are the most important technical factors that influence the quality of the soft-copy display [4]. Previous studies that evaluated the performance of soft-copy interpretation therefore focused on the impact of matrix size, magnification, and monitor luminance or on the comparative diagnostic performance of soft- and hard-copy evaluation [5–7].
Compared with these aspects, other factors with potential impact on image quality of soft-copy display, such as ambient lighting, appear relatively underappreciated. Monitor display is more vulnerable to ambient light conditions than are viewboxes with respect to reflection of light and the resulting decrease of displayed contrast ratio. Therefore, it has been recommended that ambient lighting should be less than 100 lux for optimum soft-copy display [8].
However, implementation of soft-copy displays in an external location, such as on a clinical ward, may inevitably be associated with substantially higher background lighting. In those situations, at least for specific imaging tasks, online adjustment of display brightness and contrast (window level and width) may be used to improve image contrast.
We selected catheter fragments as test structures because checking for correct catheter placement is one of the most frequent imaging tasks for chest radiography in an intensive care unit. Catheter placement requires immediate verification, and the intensive care unit would benefit from instantaneous availability of the image.
The purpose of our study was to evaluate the impact of ambient lighting on the detectability of catheter fragments using a color cathode-ray tube monitor and to test the ability of interactive window adjustment to compensate for suboptimal lighting conditions.
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
The 10 study patients had a mean weight of 85 ± 11 kg. Seven patients were receiving mechanical ventilation. Five patients had cardiac insufficiency, six patients had pleural effusions, and three patients presented with pulmonary opacifications due to either atelectasis or pneumonic infiltration.
The radiographs were taken with the patient in the supine position with a focus-film distance of 100 cm and no automatic exposure control (125 kVp, tube load ranging from 1.25 to 2.2 mAs, antiscatter grid with 40 lines per centimeter, and a grid ratio of 10:1; Mobilett II, Siemens, Erlangen, Germany).
A total of 131 low-contrast catheter fragments were superimposed on the chest radiographs in irregular orientation and distribution. The number of catheter fragments per image varied between 12 and 14. A previous hard copy of a chest radiograph of a specific patient served as a carrier for the various catheter fragments. Catheter fragments were equally distributed to high-attenuation (e.g., mediastinum and retrocardiac area) and low-attenuation (e.g., lungs) areas on the image (Fig. 1A, 1B). To superimpose the test structures on the radiograph, the carrier film with the fragments was placed on the front side of the cassette during image acquisition.
|
|
Three types of catheters were used: a central venous catheter (Howes Multi Lumen, Arrow International, Reading, PA), a pleural drain (Pleurocath, Plastimed, Le Plessis Bouchard, France), and a feeding tube (Wiruthan, Rüsch, Kernen, Germany).
Each catheter was cut into sections of 9–25 mm in length. The diameters of the catheters were 2.4 mm (central venous catheter), 2.7 mm (pleural drain), and 4.8 mm (feeding tube). None of the catheter fragments contained high-density markings that would potentially obscure underlying clinically relevant image information.
There were 64 fragments of central venous catheters, 32 fragments of pleural drains, and 35 fragments of feeding tubes. Of the 131 fragments, 88 (43 central venous catheters, 25 feeding tubes, 20 pleural drains) were superimposed over high-attenuation areas of the thorax, such as the mediastinum, retrocardiac space, or the chest wall. The other 43 fragments (21 central venous catheters, 10 feeding tubes, 12 pleural drains) were positioned over low-attenuation areas of the lung.
Image Display and Processing
Images were processed using an adaptive unsharp mask–filtering
algorithm. The parameter set comprised a sigmoidal gradation curve, a slightly
unsharp mask filtering with a weighting factor of 0.5, and a frequency rank of
0. An additional mode for dynamic range compression was applied as offered by
the manufacturer. This processing was proven to be advantageous in previous
studies and is the processing algorithm routinely used at our institution for
bedside and upright storage phosphor chest radiographs
[9]. This algorithm is designed
to approximate conventional image appearance with an improved transparency in
high-attenuation areas of the mediastinal and retrocardiac area. The
relatively large kernel for frequency filtering combined with the low
enhancement factor is thought to increase local structural contrast without
enhancement of image noise.
Images were displayed on a color display cathode-ray tube monitor (MCD 402, Philips; 21-inch [53 cm]; high-contrast color display with 1280 x 1024 matrix; standard brightness of 45 fL; color temperature, 9300°K; dot pitch, 0.25 mm; light transmission, 43%; antireflection, antistatic screen coating). The viewing station allowed a mouse-driven interactive adjustment of the window (contrast) and level (brightness) settings of the display.
Window level and window width could be altered either simultaneously or separately as warranted. Automatic restoration of standardized window parameters was possible. No other processing tools, such as magnification or spatial frequency enhancement, were available.
Interpretation Methodology
Five observers with various experience in thoracic radiology and various
familiarity with soft-copy interpretation evaluated the images. Three
observers were residents in their second, third, and fourth year. Two
observers were board-certified radiologists and specialized in thoracic
radiology. All observers were experienced in interpreting soft-copy displays.
The images were evaluated in two interpretation sessions; one session was held
under bright light and one under subdued lighting conditions, with an interval
of at least 2 weeks. To compensate for learning effects or interpretation
order bias, we changed the order of lighting conditions and the image order
for each observer.
The detectability of catheter fragments was assessed under conditions of subdued and bright ambient light. Bright ambient light consisted of overhead lighting (223 lux), whereas under subdued conditions all overhead lights were shut off (< 10 lux). The situation was approximately that of daylight on a clinical ward. The overhead lights were located in the center of the room and yielded a homogeneous increase of room lighting. Care was taken that the light source was not directly reflected by the monitor's front screen and had no blinding effect on the observers.
Before each interpretation session, observers saw three cases to become familiar with the interpretation procedure and to adapt their eyes to the lighting conditions. These extreme lighting conditions were chosen to assess the detection performance under ideal lighting conditions and under extreme suboptimal lighting conditions as would be encountered in an intensive care unit.
The observers were asked to mark the localized catheter fragments on a transparent foil that was superimposed over the monitor display. The transparent foil was taped onto the monitor in such a way that it could be easily removed but also could be manually repositioned in an identical manner. Observers did not know the exact number of catheter fragments per image. They knew, however, that each image contained multiple fragments and that the anatomic distribution of fragments per radiograph was approximately equal.
Under both lighting conditions, observers evaluated the images with and without the availability of interactive adjustment of window width and level. The evaluation was done in a standardized order. First, the interpreters marked with pen all catheter fragments they could localize without adjusting the window setting. Initially, window width and level of the soft-copy display were set to simultaneously display both high- and low-attenuation areas of the chest with optimum density and contrast.
Subsequently, the observers marked with a different color those fragments that were detected with the addition of a window adjustment. Interpreted data were separately transferred from the transparent foils to a dedicated documentation sheet. The time allowed for interpretation of each image was unrestricted. Viewing time without retrieval time was clocked by a stopwatch and recorded for each image separately with and without image manipulation.
Statistical Analysis
The interpreters' performances with and without image manipulation under
the two lighting conditions were assessed for sensitivity and false-positive
rates on a patient-by-patient basis and on an interpreter-by-interpreter
basis. A two-way analysis of variance was applied to test the significance of
the effect of ambient light and window setting adjustment on detection
performance. The level of significance was set at a p value of
0.05.
Detection rates for the subgroups of catheters with respect to their localization (low-vs high-attenuation areas) and with respect to catheter types (feeding tube, central-venous catheter, and pleural drain) were calculated by pooling the data over all observers.
Significance of difference among interpreting times was also evaluated using a two-way analysis of variance, with significance set at a p value of 0.05.
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
|
Adjusting window settings significantly increased the average detection rate from 50.2% to 58.2% (p < 0.05). The increase amounted to 8.3% under subdued light and to 7.7% under bright light (Table 2). With window setting adjustment, the difference between the detection rates under subdued light (60.8%) and under bright light (55.6%), respectively, did not reach significance (Table 2).
|
No significantly different number of false-positive interpretations was seen for the interpretations under dark or under bright background light (148 vs 151 false-positive calls). The false-positive rates were 27% and 29%, respectively. The percentage of false-positive calls that occurred with the application of window setting adjustments amounted to 39.9% (59/148) under subdued and to 39.7% (60/151) under bright ambient light.
Impact of Background Density (Low-vs High-Attenuation Area)
No significant difference was seen for the detection rates between high-
and low-attenuation areas (56.8% vs 57.9%;
Table 3). Adjusting window
settings significantly improved the detection rate in the low-attenuation
areas of the lung under bright light (46% vs 55.8%, p < 0.05).
Adjusting window settings also improved the detection rate in the
high-attenuation areas; however, differences did not reach statistical
significance.
|
Impact of Catheter Type
The pooled detection rate was best (p < 0.05) for fragments of
central venous catheters (78.1%), followed by the rate for fragments of
feeding tubes (52.6%) and pleural drains (24.4%)
(Table 4). All catheter types
were seen better under subdued light and with adjusted window settings;
however, differences did not reach statistical significance.
|
Interpretation Time
Interpretation time was longer under bright light (3.4 ± 0.9 vs 3.1
± 0.8 min); however, the difference did not reach statistical
significance. The mean time for the assessment without adjusting window
settings per patient was almost equivalent under bright and under low
background lighting (1.50 ± 0.3 vs 1.51 ± 0.4 min). Four of the
five interpreters needed a longer time for adjusting window settings under
bright lighting than under subdued lighting (mean, 1.8 ± 0.7 vs 1.6
± 0.7 min).
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Today, most viewing stations designed for diagnostic reviewing are equipped with high-resolution monochromatic black-and-white cathode-ray tube monitors. However, for financial and historical reasons, displays for Web-based or remote viewing by clinicians (e.g., on a ward or in an intensive care unit) are frequently based on the cheaper type of color monitor. Color cathode-ray tube monitors usually have less contrast, less brightness, and less spatial resolution than monochromatic cathode-ray tube monitors. If adequate image quality and sufficient information for dedicated diagnostic indications can be achieved on a color cathode-ray tube, the startup costs required for PACS (picture archiving and communication system) use would decrease significantly.
One previous study found that intrapulmonary nodules and subtle interstitial disease were equivalently visible on monochromatic or color cathode-ray tube monitors when window adjustment and zooming were available to the interpreters [10]. We therefore decided to use a color cathode-ray tube monitor in this study because it is likely that this monitor type is used for a frequently applied soft-copy display in an intensive care unit. We decided to test the specific task of localizing catheter fragments, which is not only one of the most frequent indications for acquiring a chest radiograph on the intensive care unit but also represents a diagnostic question that would benefit from instantaneously delivered information on a PACS.
Although identification of catheter lines is an important clinical task in interpreting bedside radiographs, lines are not manufactured to maximize their visibility in radiographic images, and correct localization of catheter material can be challenging. A previous study found that, at the same acquisition dose (using a 600-speed system), the visibility of support lines and tubes was inferior on computed radiography hard copies using ST IIIN plates (Fuji, Tokyo, Japan) compared with film if no image processing was performed. However, the computed radiography images significantly exceeded the performance of film after appropriate processing [11].
Factors that determine monitor performance include monitor resolution, bit depth, dot pitch luminance, and display size [12]. Most previous studies that evaluated the performance of soft-copy interpretation therefore focused on the impact of matrix size [5], magnification [5], and monitor luminance [5, 6] or on the comparative diagnostic performance of soft- or hard-copy evaluation [7, 13].
Background lighting, however, has an important impact on the contrast of a soft-copy display. Because light conditions vary considerably among various installations, the influence of background lighting in clinical practice is likely to be underestimated. With greater ambient lighting, the minimal display luminance takes on higher values, and the maximal luminance must be increased accordingly to maintain a constant contrast ratio. An increase in the maximal luminance of a cathode-ray tube display, however, is limited by the fact that the spot beam widens for higher beam currents, which decreases spatial resolution [8].
The objectives of our study were to quantify the negative impact of background lighting on the detection of catheters with soft-copy displays of bedside chest radiographs and to evaluate whether interactive window and level adjustment would compensate for these negative effects.
The detectability of catheters was generally higher under subdued light than under bright ambient light. This was true for all observers, regardless of their level of experience. The difference between mean detection rates (51.8% vs 56.6%) reached statistical significance. Detection performance under subdued light but without window adjustment was inferior to the detection rate under bright light with window adjustment (52.5% vs 55.6%). Although the difference did not reach statistical significance, it indicates the importance of window setting adjustment for this particular imaging task. This finding is further supported by the fact that the difference between the detection rates under bright light with window adjustments (55.6%) and under ideal conditions of subdued ambient light with window adjustments (60.8%) also did not reach statistical significance. This was true for all three types of catheters. Adjusting window settings appears to largely compensate for suboptimal lighting conditions and to make even suboptimal lighting conditions acceptable for the specific task of localizing catheter material.
The fact that we could not find a statistically significant difference in performance under different lighting conditions when window setting adjustments were applied does not mean that performance really reaches equivalence under different light conditions. The deviations among interpreters' performances in combination with the relatively small number of interpreters may have contributed to weakening the power of our statistical test. On the other hand, considering that data were acquired under extreme conditions with respect to low lesion conspicuity and suboptimal lighting conditions, our data are suited to show the strength of adjusting window settings for compensation of contrast loss caused by light reflection. All interpreters saw significantly more catheter fragments with interactive window adjustment under both lighting conditions.
The gain in detection rate was more pronounced under subdued light than under bright ambient light (Table 2). The fact that on-line window adjustment did not achieve the same effect in bright lighting as in low lighting suggests that the loss of contrast ratio of the monitor under bright ambient lighting cannot be fully compensated by window adjustment.
No significantly different detection rate was seen for catheters superimposed over the mediastinum, the retrocardiac region, or the chest wall compared with those located in low-attenuation areas of nonobscured lung (56.8% vs 57.9%). Window adjustment was found to be equally advantageous in both areas.
The role of individual window adjustment remains somewhat controversial and may also depend on the lesion type. In a previous study, the accuracy for the diagnosis of intrapulmonary nodules with soft-copy interpretation decreased with window manipulation because of a lower sensitivity and an increased false-positive rate [14]. Another study, however, reported an increased visualization of inserted soft-tissue foreign bodies (e.g., plastic, glass, graphite, and wood in radiographs of a cadaveric hand) with soft-copy interpretation including window adjustment compared with film and digital hard copies [15]. The authors reported comparable viewing times; no comment was made about the false-positive rates. In our study, we did not find an increased false-positive rate with window adjustments, which was true for both lighting conditions.
The rather high rate of false-positive findings in our study may be due to the fact that the catheter fragments were relatively short and also were placed in unusual anatomic locations. Therefore, observers may have misinterpreted parenchymal densities as catheter fragments.
We were interested in quantifying the effect of adjusting window settings as a function of background lighting. Therefore, we did not assess the effect of adjusting window settings in separate interpretation sessions but rather determined the number of additional catheter fragments seen when window setting adjustments were available. We decided to test rather extreme lighting conditions to assess the range of performance under ideal and extremely suboptimal conditions. We tried to reduce learning effects and interpretation order bias by altering the order of images for each observer and by altering the order of lighting conditions. Although learning effects cannot be completely excluded, we consider these effects to be small because observers were asked to perform a specific detection task that did not require an intellectual diagnostic process.
The number of clinical images (10) was rather small. However, these images served only as anatomic background for localizing catheter fragments. Patients were selected in a way that would be representative of a spectrum of body constitution and underlying disease. In addition, catheter fragments were placed in various anatomic areas. Although for complex diagnostic evaluations more images would certainly be necessary, we think that for this specific interpretation task the number of images was sufficient. The study setup was chosen so that all images were clinically indicated and of adequate quality for clinically relevant diagnostic purposes.
Overall sensitivity for detecting the catheter fragments was rather low (51.8–58.2%), indicating that the fragments were subtle and consequently difficult to detect even under ideal subdued lighting conditions. Our study design was aimed to find even small performance differences that can be difficult to prove if the detection task is too obvious. Because it is diagnostically important not only to assess the course of the catheter but also to localize its termination, we used only short fragments (range, 9–25 mm) to simulate the tip of a catheter. In fact, most fragments that we used actually were the terminal 20 mm of a catheter. Only low-contrast types of catheters were used that did not contain an indicating high-contrast stripe. We expect that the detection rate would be higher in clinical practice for catheters that are imaged in full length because of the higher conspicuity of a longer catheter course. However, with respect to the localization of the catheter tip, similar imaging constraints based on the signal-to-noise ratio and the structural contrast are valid also in clinical images as simulated in our study.
It is true that structures superimposed over an anatomic background underlie different scatter conditions than structures lying within the body. We tried to minimize these effects by taping the fragments on a carrier that was placed directly onto the cassette underneath the patient's body. The resulting appearance of the catheters in the image was very realistic, as shown in Figure 1A, which shows both a real central venous catheter and superimposed catheter fragments.
Both acquisition parameters and the processing algorithm as applied in this study were designed to increase the transparency of high-attenuation areas of the mediastinum and to increase local structural contrast, thereby supporting the detectability of catheter fragments. The exposure conditions stated (125 kVp, with grid) are standard at our institution. Because all images were part of the diagnostic workup of the patients, images were obtained in the usual fashion and underwent the regular reviewing process. The dynamic range compression mode is used to reduce dynamic range and is also part of the routine algorithm.
The detection rate of catheters would probably be different with other parameters. However, detection performance would still be subject to ambient lighting and online processing. So even though the absolute detection rates might change with different acquisition techniques or data processing, the relative impact of ambient lighting will remain.
Also, for conventional radiology extraneous light has been shown to decrease detectability of low-contrast objects [16, 17]. Three studies in the literature have evaluated the effects of background lighting on the performance of digital images and have reported somewhat controversial results depending on whether hard-copy or soft-copy interpretation was applied. One of these studies tested the detection of artificial lesions on hard copies of bitewing radiographs of extracted teeth and reported significant differences as a function of observers, detector types, and lesion sizes, but no significant impact of background lighting [12]. In opposition to that, results of the other two studies suggested a significant decrease of detection performance with soft-copy interpretation as a function of increased background lighting. One study referred to the gray-scale perception using a standardized test pattern (Society of Motion Pictures and Television Engineers, White Plains, NY) [18]. The second study tested the detectability of osseous lesions [19]. None of the previous reports evaluated the effect of window setting on the detection performance.
Images displayed on a monitor are viewed differently than images displayed on viewboxes [20]. It has been reported that approximately 20% of interpretation time is spent on image processing functions. As a result, overall viewing time was significantly shorter for the computed radiography film than for the soft-copy evaluation (64.5 ± 24.7 vs 91.2 ± 47.6 sec, p < 0.05). The authors of that study suggested that interpretation time might be dramatically reduced if workstations were designed so that individual radiologists could determine their own settings. Another possibility could be to predefine processing settings according to specific imaging tasks (e.g., for structural enhancement in the mediastinal or in the pulmonary area). Nodine et al. [21] presented an elaborate soft-copy processing algorithm for optimal display of catheter lines that included a widening of the displayed dynamic range combined with a spatial filtering process for local contrast optimization. With this algorithm, they found a significantly improved detection rate for catheter fragments on soft-copy display, especially for less experienced interpreters.
Our results primarily pertain to the specific type of monitor we tested. Results may vary with other monitor types with respect to the quantities we describe; however, the conclusion that adjusting window settings provides beneficial effects for the localization of catheters under suboptimal ambient light conditions probably holds true for other monitor types as well. Cederberg et al. [12] could not find significantly different performances for the detection of artificial enamel lesions in extracted teeth using four types of cathode-ray tube monitors with slightly different technical parameters, which suggests comparable image properties of various types of cathode-ray tubes.
The most recently introduced active matrix liquid crystal displays offer some organizational, financial, and display advantages compared with the traditional curved-surface cathode-ray tube monitors such as a higher luminance, a shorter depth of the monitor with a lower weight, and a flat-panel display [22, 23]. Another advantage of the flat-surface liquid crystal display is the reduction in glare from reflections when ambient light is increased. One study reported a superior ability to discriminate gray-scale differences with the liquid crystal display compared with the cathode-ray tube display under bright ambient lighting [18]. Results at our institution showed an equivalent detection performance of both displays under subdued and bright lighting conditions. Although both displays showed a decreased detection rate under bright lighting, results showed a significantly lower diagnostic confidence for the cathode-ray tube than for the liquid crystal display (Scharitzer et al., unpublished data).
We conclude that for this particular type of monitor display, interactive window and level adjustment of the soft-copy display largely compensated for the impact of bright ambient light. The detection rates under ideal subdued and under suboptimal bright light conditions when window setting adjustments were applied did not reach statistical significance. Whether these results can be transferred to diagnostic tasks other than the detection of catheter material remains to be seen.
Although the vulnerability of soft-copy display to ambient light is different for different monitor types, the problem of reduced contrast resulting from suboptimal lighting conditions remains valid for soft-copy display in general. The use of interactive window setting adjustments should be encouraged even under seemingly ideal light conditions.
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
CiteULike 
Complore 
Connotea 
Del.icio.us 
Digg 
Reddit 
Technorati What's this?
This article has been cited by other articles:
|
|
 |
|
  D S Brettle Display considerations for hospital-wide viewing of soft copy images Br. J. Radiol., July 1, 2007; 80(955): 503 - 507. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
D P Thompson, C J Koller, and J P Eatough Practical assessment of the display performance of radiology workstations Br. J. Radiol., April 1, 2007; 80(952): 256 - 260. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 P. C. Brennan, M. McEntee, M. Evanoff, P. Phillips, W. T. O'Connor, and D. J. Manning Ambient Lighting: Effect of Illumination on Soft-Copy Viewing of Radiographs of the Wrist Am. J. Roentgenol., February 1, 2007; 188(2): W177 - W180. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
