British Journal of Radiology (2006) 79, 425-431
© 2006 British Institute of Radiology
doi: 10.1259/bjr/28579947
Optical density variations in CT films and their effect on image quality
I A Tsalafoutas, PhD1,
G V Papoutsis, RT2,
P N Maniatis, MD3 and
K A Gogos, MSc4
1 Medical Physics Unit, 'Konstantopoulio - Agia Olga' Hospital, 3-5 Agias Olgas, Nea Ionia, 142 33, Athens, 2 Field Service Engineering Department, General Electric Medical Systems, 156 Cyprou Avenue and 91 Konstantinoupoleos Str, Argyroupolis, 164 51, Athens, 3 Computed Tomography Department, 'Konstantopoulio - Agia Olga' Hospital, 3-5 Agias Olgas, Nea Ionia, 142 33, Athens, 4 Nuclear Medicine Department, 'Hygeia' Hospital, 4 Erythrou Stavrou, 151 23, Maroussi, Athens, Greece
 |
Abstract
|
|---|
It was recently reported that optical density (OD) variations were observed in CT films printed with a laser camera, depending on the printing format and the frame position within the film. The purpose of the present study was to investigate if these variations are common to both laser and dry-film printers and if the different OD settings along with day-to-day and frame-to-frame variations may affect the image quality. Eight laser and five dry-film printers installed at 12 different CT facilities were tested. For each one, the SMPTE test pattern was printed on all frames of a film using the same printing format. The ODs of the 0%, 10%, 40% and 70% patches of the 11-step greyscale of the SMPTE patterns were measured with a densitometer in all frames, while all films were examined on a viewing box to assess subjectively the image quality by visual inspection of the test pattern. A wide range of OD settings and variations were recorded. Frame-to-frame variations in the same film of up to 0.19, 0.15 and 0.21 OD, were observed for contrast index (CI, the OD difference of patches 10% and 70%), speed index (SI, the OD of patch 40%) and maximum OD (ODmax, the OD of patch 0%), respectively. The variations were not always of the same magnitude, nor always followed the same pattern, even for printers of the same model. Considering all films and frames, the CI ranged from 1.26 to 1.74, the SI from 0.68 to 1.43 and the ODmax from 2.5 to 3.11 OD, well beyond the proposed settings and tolerances of 1.55±0.15, 1.15±0.1 and 2.45±0.1 given in the literature for CI, SI and ODmax, respectively. Despite these large differences, the various problems that were identified in image quality from the visual inspection of the films could not be directly attributed to OD settings, as films with similar CI, SI and ODmax presented quite different image quality levels. Therefore, for routine quality control, thorough visual inspection of the SMPTE test pattern provides all the necessary information about the imaging chain status.
|
Introduction
|
|---|
CT was first introduced into clinical practice in the early 1970s and since then has become one of the most useful diagnostic tools. The technology of CT scanners is continuously and rapidly evolving while significant developments have also occurred in CT film printers, as laser and dry printers superseded CRT based multiformat cameras (hard-copy cameras).
Nowadays, in CT departments, diagnosis is more and more frequently performed on workstation monitors. However, film still remains an important means for diagnosing, storing and transmitting CT images. Indeed, whilst monitors and films are both currently used for the interpretation of CT images, in many cases diagnosis is based solely on film reading. Since for viewing a CT image, the window width (WW) and window level (WL) are adjusted so that the structures of interest are best imaged on the monitor, it is consequently important to ensure that the image characteristics are maintained when printed on film.
In order to test medical imaging systems and hardcopy devices, the Society of Motion Picture and Television Engineers (SMPTE) formed a committee that developed and tested a number of patterns, until they agreed on the final version of the SMPTE monochrome test pattern that provides both qualitative and quantitative information [1]. The characteristics of SMPTE pattern, its use for acceptance and quality control (QC) purposes and relevant technical details on the design of film printers are summarized in the report of the American Association of Physicists in Medicine (AAPM) Diagnostic X-ray Imaging Committee Task Group No 1 [2].
In the AAPM report [2], reference values are given for video and laser cameras concerning the optical densities (OD) that certain patches of the 11-step greyscale of the SMPTE pattern should have when printed on film, along with the OD variation tolerances for the QC of these systems. However, in a recent study [3] it has been reported that the ODs of these patches vary with printing format and with frame position within the same film.
In the present study, the frame-to-frame OD variations within the same film were further investigated for a number of laser and dry printers, to identify if these variations are dependent on the type of film printer and whether being combined with different OD settings and day-to-day variations can affect the image quality in any way.
|
Materials and methods
|
|---|
In this study, 12 different CT facilities were included; 7 equipped with a laser and 4 with a dry-film printing device, while in one facility two film printers (one laser and one dry) were interfaced to the same CT scanner. All CT scanners were manufactured by the same company (GE Medical Systems, Milwaukee, WI), while film printers were manufactured by Agfa (Agfa-Gevaert A.G., München, Germany) and Kodak (Eastman Kodak Company, New York, NY).
In each facility, the SMPTE test pattern was viewed on the CT monitor, setting the WW to 100 and the WL to 0 (except for the Hispeed models where the WL was set to 1024 automatically when the SMPTE was imaged) [2]. The SMPTE test pattern is shown in Figure 1
and a detailed description of its characteristics is given in the AAPM report [2]. According to the suggestions of the AAPM report, the brightness and contrast of the TV monitor can be considered adequately adjusted when the 5% and 95% average picture level patches inset, respectively, in the 0% and 100% patches (located at the two ends of the 11-step greyscale) and the low and high contrast bar-patterns (located at the centre and the four corners of the SMPTE test pattern) are clearly discernible. In all TV monitors examined, these criteria were fulfilled and the adjustment of contrast or brightness was not considered necessary.
Subsequently, and without changing the window settings, the SMPTE pattern was printed so as to cover the 20 frames of a film (5 rows by 4 columns). The OD values of the 0%, 10%, 40% and 70% patches of the 11-step greyscale SMPTE patterns of all frames were measured with a calibrated optical densitometer (RMI 331 Densitometer; X-Rite, Grandville, MI). The OD of patch 40% is defined as the speed index (SI), the OD difference of patches 10% and 70% as the contrast index (CI), while ODmax is the OD of patch 0% [2].
In two CT facilities, the above procedure was repeated a few months after the first test in order to identify possible day-to-day variations. In one CT facility, the test procedure was carried out three times with the same laser camera and three different film-types (after applying the necessary sensitometric adjustments), while for one film type the test was also repeated a few months after the first test. In the latter facility, in order to adjust the photographic processor after film changes, the procedure described in the user manual of the laser camera was carried out. The laser camera produced a 21-step greyscale setting the target density at the value of 3.3 OD, as proposed by the manufacturer. The maximum OD (the OD of the 21st step) was measured with the optical densitometer and was manually fed into the camera. This procedure was repeated until a film with maximum OD within ±0.05 of the target density was produced. Next the ODs of all steps were measured and manually fed into the laser camera for updating the look-up tables [2].
In addition to the OD measurements, all the films were observed on the same viewing box to assess the qualitative characteristics of the printed SMPTE patterns. Two observers graded in consensus the images for contrast (in both low and high ODs) and resolution (in both low and high contrast) using an arbitrary five-point scale ranging from 1 (not discernible) to 5 (perfectly discernible) and commented on the presence of any type of artefacts. For contrast grading, the visibility of the 5% and 95% insets to the 0% and 100% patches, respectively, was assessed. For low contrast resolution grading, the visibility of the low contrast bar-patterns was assessed; similarly, for high contrast resolution grading, the high contrast bar-patterns located beside the respective low contrast bar-patterns were evaluated.
|
Results
|
|---|
Overall, 18 films acquired from 13 different printers were examined. The mean values and the maximum variation (max–min) of the CI, SI and ODmax recorded in the 20 frames of each film are given in Table 1, along with the results of the qualitative assessment. The tests with decimal numbers indicate those tests made on the same film printer, in order to investigate the impact of film changes and day-to-day variations on frame-to-frame OD variations and image quality.
View this table:
[in this window]
[in a new window]
|
Table 1. Results of the optical density(OD) measurements and visual assessment of image quality in the 18 films studied. The mean values of contrast index (CI), speed index (SI) and maximum OD (ODmax), and the maximum frame-to-frame differences (max–min) observed are shown. Mean values outside the ranges (1.55±0.15, 1.15±0.1 and 2.45±0.1 for CI, SI and ODmax, respectively) proposed for the quality control (QC) of a laser camera [2], are given in italics. The maximum frame-to-frame differences observed for laser (tests 8.1–8.2) and dry printers (tests 9–13) are given in bold. The contrast grading in the high and low ODs (around the ODs of inset patches 5% and 95%, respectively) and the low contrast (LC) and high contrast (HC) resolution grading are shown, as well as comments on the presence of artefacts
|
|
Most of the observed OD variation patterns are given in Figures 2–6. The variation patterns depicted in Figure 2 refer to the same laser printer (Scopix LR 3300) for 4 tests made with 3 different films. The corresponding sensitometric curves (used for updating the look-up tables) are depicted in Figure 7. In Figure 3, the OD variation patterns of 6 laser printers of the same model (Scopix LR 3300) are given (for printers 1 and 2, the results from the tests that exhibited the largest variations are depicted). In Figure 4, the OD variation patterns of 3 different laser printers (Scopix LR 3300, Scopix LR 5200, Ektascan 2180) are given (for Scopix LR 3300, the results from the test that exhibited the largest variations is depicted). In Figure 5, the OD variation patterns of 5 dry printers (three DryView 8100, one DryView 8700 and one Drystar 3000) are given. Finally, in Figure 6, the OD variation patterns of the laser and the dry printers that exhibited the largest frame-to-frame variations (one Scopix LR 3300 and one DryView 8100, respectively) are given for comparison.
View larger version (21K):
[in this window]
[in a new window]
|
Figure 2. The variation of the contrast index(CI), speed index (SI) and maximum optical density (ODmax) values with frame numbers for tests 1.1 (Agfa film), 1.2 (Ferrania film), 1.3 and 1.4 (Konica film) performed on different days with the same laser printer (Scopix LR3300, Agfa) at the same CT installation. The frames have been numbered starting from the upper left corner, and vertical lines indicate the start of each of the 5 rows with 4 frames each.
|
|
View larger version (23K):
[in this window]
[in a new window]
|
Figure 3. The variation of the contrast index(CI), speed index (SI) and maximum optical density (ODmax) values with frame numbers for tests performed at different CT installations which all had the same laser printer model (Scopix LR3300, Agfa) and used the same film (LT 2B daylight, Agfa).
|
|
View larger version (19K):
[in this window]
[in a new window]
|
Figure 4. The variation of the contrast index(CI), speed index (SI) and maximum optical density (ODmax) values with frame numbers for tests performed at different CT installations, using three laser printer models (3: Scopix LR3300-Agfa; 7.1: Scopix LR5200-Agfa; 8.1: Ektascan 2180-Kodak).
|
|
View larger version (22K):
[in this window]
[in a new window]
|
Figure 5. The variation of the contrast index(CI), speed index (SI) and maximum optical density (ODmax) values with frame numbers for tests performed at different CT installations, using different dry printer models (9,10,11: DryView 8100-Kodak; 12: DryView 8700-Kodak; 13: Drystar 3000-Agfa).
|
|
View larger version (15K):
[in this window]
[in a new window]
|
Figure 6. The variation of the contrast index(CI), speed index (SI) and maximum optical density (ODmax) values for the laser and dry printers that exhibited the largest frame-to-frame variations (3: Scopix LR3300-Agfa; 9: DryView 8100-Kodak).
|
|
View larger version (15K):
[in this window]
[in a new window]
|
Figure 7. The sensitometric curves used for updating the look-up tables before acquiring the SMPTE films (tests 1.1 to 1.4). The dashed line is the sensitometric curve that resulted when the film changed from Agfa to Ferrania before updating the look-up tables. The latter curve is included in the graph to demonstrate the influence of look-up tables on the sensitometric characteristics of CT films.
|
|
The OD variation patterns of CI, SI and ODmax given in Figure 2
present some similarities. All these parameters obtain their maximum values in the upper left frame and then, when moving to the right, decrease to a minimum (at the 3rd frame) and then increase again. This pattern is repeated for each row and in some cases a slight decrease from the first to the last row is also observed. The mean values of CI, SI and ODmax are quite similar even if the respective sensitometric curves are quite different. However, the maximum variations in CI and ODmax ranged from 0.07 to 0.15 and 0.06 to 0.16, respectively, with no obvious correlation with the sensitometric curves. It is important to note that when the film was changed from Agfa to Ferrania, the sensitometric curve (shown in Figure 7 with the dashed line) was significantly faster than that of Agfa, obtaining a maximum OD of 3.72 instead of the 3.3 target OD. After the look-up tables were updated, the sensitometric curve of the Ferrania film was found to be significantly slower than that of Agfa film. However, the CI, SI and ODmax were almost equal to those obtained with the Agfa film.
It can also be deduced from the tests repeated with the same film (tests 1.3–1.4, 2.1–2.2, 8.1–8.2) that day-to-day variations did not significantly change the OD variation pattern. However, differences in the maximum frame-to-frame OD variations and the mean values of CI, SI and ODmax were observed, as in the case of the Ektascan 2180 (tests 8.1–8.2) where the mean value of SI changed between tests from 1.02 to 1.39.
In Figure 3, it is apparent that not all Scopix LR3300 presented the same OD variation pattern or the same maximum variations in CI, SI and ODmax. In Figure 4, the more extreme pattern observed for Scopix LR3300 is compared with the less extreme patterns observed for Scopix LR5200 and Ektascan 2180 laser printers. In Figure 5, one can appreciate the variety of OD variation patterns observed for the various dry printers tested. Finally, in Figure 6 it is obvious that the variation patterns of the laser and dry printers that exhibited the largest CI, SI and ODmax frame-to-frame variations are completely different.
As far as the qualitative assessment of images is concerned, films with reduced contrast and resolution, and various artefacts were identified. These problems were common to all film frames and no frame-to-frame variations in image quality were observed, with one exception discussed in the end of this section.
No problems were observed in the contrast in the high ODs, as the 5% inset patch was always perfectly (grade 5) or well (grade 4) discernable from the 0% background. However, for low ODs there were three films (tests no. 5, 2.1 and 4) where the 95% inset patch was not discernable at all (grade 1) or was poorly discernable (grade 3) from the 100% background. While this could be attributed to the very low SI of all three films, for the film from test no. 2.2, which also exhibited a very low SI, the respective contrast grade was 4.
On the other hand, the deterioration of low and high contrast resolution observed in some of the films could not be associated with OD settings. While in some cases a reduced resolution could be attributed to other printer problems causing geometrical distortion and background mottle (as in tests no. 2.1, 2.2, 9 and 11), a reduced low contrast resolution was also observed in the films from tests no. 8.1 and 8.2 with no obvious reason.
Finally, it is worth mentioning that the only exception in the image quality homogeneity among frames was the film from test no. 4. In this film, alternating bands of white-black and white-grey line pairs in the high contrast bar patterns (an artefact we termed "banding") were apparent in all frames. However, this artefact was more intense in the outer left frames (the first frame of each row) and specifically in the upper left and lower left high contrast bar patterns. This was indeed the only case where a definite variation of image quality among frames was observed.
|
Discussion
|
|---|
The AAPM report [2] gives mean values and acceptable tolerances of CI, SI and ODmax as 1.55±0.15, 1.15±0.1 and 2.45±0.1, respectively. These values, however, refer to the periodic quality control (QC) of a laser camera where the same format and frame position are monitored using the Kodak SO-497 film. These serve as a reference for the accepted OD settings and day-to-day variations but are not applicable to frame-to-frame variations. For the latter variations, the AAPM report [2] recommends that OD differences of ±0.05 or slightly wider within one sheet of film may be accepted. However, the report does not state whether exceeding the proposed tolerances could have an effect on image quality. Furthermore, as has been noted in a previous study [3], limits have not been set for frame-to-frame variations indicating a certain malfunction of the film printer.
The reason for frame-to-frame variations has not yet been identified. While some possible reasons have been proposed by Agfa technicians concerning the results of our previous paper [3], in view of the variations also observed in dry printers, where different technologies are used for film exposing and processing, none seem very convincing. The other image quality problems identified in the films studied could be attributed to a number of factors (e.g. the geometrical distortion is most probably due to a worn polygon mirror) and can usually be identified and repaired by the camera manufacturer technicians, who usually utilize other types of test patterns that the cameras are able to produce. As written in the AAPM report [2], the SMPTE test pattern can be used to identify image quality problems, but not to trace the exact origin of the problem within the imaging chain. This requires further investigation by the camera (and perhaps the CT) technicians and it is outside the scope of this paper. Whilst we are eager to study the above issues further, this is not currently possible because of the lack of detailed information on design and sources of problems provided by the companies concerned.
The mean CI values given in Table 1 (except for tests 2.1, 2.2 and 4) are within the reference range. However, many of the SI and all the ODmax values given in Table 1 are outside the range of values given by the AAPM report [2]. Thus, the printers and films investigated in our study exhibit a larger variability in performance. The maximum differences in CI, SI and ODmax observed in the present study due to frame-to-frame variations reached 0.19 OD, 0.15 OD and 0.21 OD, respectively (film no. 3). Thus, if both day-to-day and frame-to-frame variations are taken into account, the CI, SI and ODmax on any frame for a specific film printer could well vary within ±0.25, ±0.2 and ±0.2, respectively, from the central OD settings. It must be also noted that considering all the frames of all films, the CI ranged from 1.26 to 1.74 (0.48 OD difference), the SI from 0.68 to 1.43 (0.75 OD difference) and the ODmax from 2.5 to 3.11 (0.61 OD difference). Despite these large OD variations, image quality as assessed by visual inspection of the SMPTE test pattern could not be correlated with the OD values.
The results of this study confirmed that frame-to-frame OD variations were common among the different printers tested, though were not always of the same magnitude or pattern. However, an OD variation large enough to reduce image quality only for certain frames of the same film was not identified. Therefore, whatever the frame-to-frame differences between ODs, if the SMPTE test pattern characteristics remain clearly and equally discernable in all frames during visual inspection, it should be safe to assume that the quality of printed clinical CT images remains adequate for diagnosis.
In conclusion, we propose that for a certain CT facility, once the reference values for the OD of all the patches in the 11-step grey scale of the SMPTE test pattern and the CI, SI and ODmax have been determined, as well as the frame-to-frame OD variations (at least for the most commonly used printing format), periodic visual inspection of the SMPTE test pattern on the TV monitor and on all the frames of a film should be adequate for routine QC purposes. However, when from the visual inspection a loss of image quality is suspected, measurements of the ODs in a film fully covered with the SMPTE test pattern should be carried out to determine whether the camera should be re-adjusted or repaired.
Received for publication May 4, 2005.
Revision received September 9, 2005.
Accepted for publication September 13, 2005.
|
References
|
|---|
- Specifications for medical diagnostic imaging test pattern for television monitors and hard-copy recording cameras. SMPTE Recommended Practice 1986, RP 133-1986. SMPTE Journal 1986;95:693–5.
- Gray JE, Anderson WF, Shaw CC, Shepard J, Zeremba LA, Lin PP. Multiformat video and laser cameras: history, design considerations, acceptance testing and quality control. Report of AAPM Diagnostic X-ray Imaging Committee Task Group No1. Med Phys 1993;20:427–38.[Medline]
- Tsalafoutas IA, Tsapaki V, Koulentianos E, Triantopoulou C. Quality control of a laser camera with the SMPTE test pattern: optical density variations with printing format and frame position. Br J Radiol 2004;77:52–6.[Abstract/Free Full Text]
This article has been cited by other articles:
|
|
|
|
 
K. A. KHALIFAH, A. BRINDHABAN, G. A. BALOUL, F. A. BATHER, and M. ABDULGAFOOR
Optical Density Changes In Dry-processed Films
Radiol. Technol.,
September 1, 2007;
79(1):
9 - 16.
[Abstract]
[Full Text]
[PDF]
|
|
|
Top
Abstract
Introduction
