DOI:10.2214/AJR.07.3740
AJR 2008; 191:W181-W189
© American Roentgen Ray Society
Influence of Tagged Fecal Material on Detectability of Colorectal Polyps at CT: Phantom Study
Ayso H. de Vries1,
Henk W. Venema1,2,
Jasper Florie1,
Chung Y. Nio1 and
Jaap Stoker1
1 Department of Radiology, Academic Medical Center, University of Amsterdam,
Meibergdreef 9, 1105 AZ Amsterdam, The Netherlands.
2 Department of Medical Physics, Academic Medical Center, University of
Amsterdam, Amsterdam, The Netherlands.
Received January 28, 2008;
accepted after revision April 11, 2008.
Address correspondence to A. H. de Vries
(Ayso.devries{at}amc.uva.nl).
A grant was received from Philips Healthcare, which was not involved in
designing or conducting the study and did not have access to the data. Philips
Healthcare was not involved in analyzing the data or writing the manuscript
and was not asked for approval.
WEB
This is a Web exclusive article.
Abstract
OBJECTIVE. The purpose of this study was to determine the influence
of tagged material on the minimal radiation dose needed to detect colorectal
polyps at CT.
MATERIALS AND METHODS. The study was conducted in two phases. In the
first, three experienced observers determined the visibility of sessile polyps
(6 mm) at five contrast levels (300, 480, 790, and 1,040 HU and air) and five
tube charge levels (10, 14, 20, 28, and 40 mAs) in an anthropomorphic phantom.
Each polyp was present in one of eight possible locations. The mean tube
charge threshold for 90% correct responses was determined for each contrast
level. Blinded observers performed independent 2D readings. In the second
phase of the study, three 150-cm virtual colons were evaluated at two contrast
levels (300 and 480 HU) and at five tube charge levels between 20 and 80 mAs.
The three colons contained 18 randomly located polyps. The mean tube charge
threshold for 90% sensitivity was determined for each contrast level.
RESULTS. In the first phase of the study, the estimated tube charge
thresholds for 300, 480, and 790 HU were 24.0, 16.3, and 6.2 mAs. At 1,040 HU
and in air, all polyps were detected at the lowest tube charge setting (10
mAs). In the second phase, the tube charge thresholds for 90% sensitivity at
300 and 480 HU were 70 and 35 mAs, respectively.
CONCLUSION. If polyps are covered by fecal material, a considerably
higher tube charge setting is needed for adequate visualization than is needed
for polyps in a completely cleansed colon, especially when the density of the
tagged residue is low.
Keywords: bowel preparation • CT colonography • dose reduction • fecal tagging • phantom study
Introduction
CT colonography (CTC) is a promising technique of screening for colorectal
cancer and colonic polyps. An important disadvantage of the technique is that
many patients consider the bowel preparation burdensome
[1]. Labeling fecal material
with an oral contrast agent (fecal tagging) enables patients to prepare for
CTC with less extensive bowel preparation
[2–4]. Minimizing bowel
preparation, however, may increase patient compliance
[5–7]
but results in a higher amount of residual feces in the colon
[8].
Because of reduced contrast (i.e., reduced difference in attenuation)
between the polyp and the luminal contents, polyps covered by fecal material
are less conspicuous than polyps surrounded by air, even if oral contrast
material has been administered. Figure
1A,
1B,
1C,
1D shows that the submerged
polyp is less visible than the haustral fold in air at 40 mAs. Another factor
that influences polyp conspicuity is radiation dose. Especially in low-dose
scan protocols, the noise of CT images is prominent. In Figure
1A,
1B,
1C,
1D, this phenomenon is
illustrated by the submerged polyp visible at 40 mAs but hardly visible at 5
mAs.
View larger version (125K):
[in this window]
[in a new window]
[as a PowerPoint slide]
|
Fig. 1A —Influence of tagged material on visibility. 75-year-old man
with colonic polyp. CT scan (40 mAs; window width, 1,250 HU; level, –50
HU) shows colonic wall and polyp (white arrow). Gray arrow indicates
haustral fold in transverse colon surrounded by air.
|
|
View larger version (133K):
[in this window]
[in a new window]
[as a PowerPoint slide]
|
Fig. 1B —Influence of tagged material on visibility. Simulated CT
scans at tube charges lower than in A: 20 mAs (B), 10 mAs
(C), and 5 mAs (D). White arrow indicates 6-mm polyp submerged
in fecal material in descending colon proven at colonoscopy. Polyp is highly
visible in A but hardly visible in D. In comparison, haustral
fold in transverse colon surrounded by air (gray arrow) remains
highly visible in D.
|
|
View larger version (143K):
[in this window]
[in a new window]
[as a PowerPoint slide]
|
Fig. 1C —Influence of tagged material on visibility. Simulated CT
scans at tube charges lower than in A: 20 mAs (B), 10 mAs
(C), and 5 mAs (D). White arrow indicates 6-mm polyp submerged
in fecal material in descending colon proven at colonoscopy. Polyp is highly
visible in A but hardly visible in D. In comparison, haustral
fold in transverse colon surrounded by air (gray arrow) remains
highly visible in D.
|
|
View larger version (153K):
[in this window]
[in a new window]
[as a PowerPoint slide]
|
Fig. 1D —Influence of tagged material on visibility. Simulated CT
scans at tube charges lower than in A: 20 mAs (B), 10 mAs
(C), and 5 mAs (D). White arrow indicates 6-mm polyp submerged
in fecal material in descending colon proven at colonoscopy. Polyp is highly
visible in A but hardly visible in D. In comparison, haustral
fold in transverse colon surrounded by air (gray arrow) remains
highly visible in D.
|
|
With regard to using CTC for screening, the goal should be the lowest
possible patient burden and radiation dose. To our knowledge, the influence of
the use of oral contrast material on the minimal radiation dose for
qualitatively acceptable CTC has not been determined. We performed a phantom
study to determine the magnitude of this effect.
Materials and Methods
Study Design
This phantom study consisted of two parts. The aim of the first phase was
to determine the detectability of clinically relevant polyps (
6 mm) in
tagged material of various densities or in air as a function of the tube
charge—that is, the product of tube current and exposure time divided by
the pitch—in an idealized setup. We performed the second phase to
translate the results of the first phase into clinically applicable
performance values, such as sensitivity and number of false-positive
findings.
Phantom
A cylindric water-filled polyethylene drum with a diameter of 34 cm was
used to mimic the abdomen of a fairly obese patient
(Fig. 2). We devised a
synthetic colonic segment by placing a cylindric polymethyl methacrylate
(PMMA) tube in the center of this drum. We positioned this phantom with the
long axis of the tube parallel to the long axis of the drum. The tube
contained eight PMMA rings with a thickness of 11 mm and an inner diameter of
50 mm in the order of the diameter of a distended colon segment. Each ring
could contain a PMMA hemisphere with a diameter of 6 mm, mimicking a 6-mm
sessile polyp. We chose 6 mm because it often is considered the smallest
clinically relevant size for polyps
[9]. All rings were separated
by a thin PMMA ring (inner diameter, 40 mm; thickness, 2 mm), representing an
artificial haustral fold.
View larger version (115K):
[in this window]
[in a new window]
[as a PowerPoint slide]
|
Fig. 2 —Phantom (width, 1,250 HU; level, –50 HU). CT scan shows
water-filled drum (1) with centrally placed polymethyl methacrylate cylinder
(2), representing colon. Cylinder is filled with contrast material (3).
Contrast between lumen and border is 480 HU. Filling defect represents 6-mm
polyp.
|
|
We use the terms "contrast" and "contrast level" to
indicate the difference in attenuation between polyps and contrast material or
air in the lumen. The attenuation of PMMA is approximately 120 HU. We used
PMMA because it is readily avail able and its attenuation is reasonably close
to that of soft tissue. The tube was completely filled with either contrast
material or air for each scan. The contrast material was a mixture of iodine
(sodium amidotrizoate, 300 mg I/mL, hospital pharmacy) and water. We chose
this concentration of contrast material so that the differences in attenuation
between the polyp and the lumen would be approximately 300, 500, 800, and
1,000 HU. These attenuation values encompassed the range of contrast measured
in tagged CTC examinations performed at our hospital with a regimen of 4 L of
polyethylene glycol and 200 mL of 300 mg I/mL of meglumine ioxithalamate
(Telebrix, Guerbet). Other regimens with similar attenuation have been
described [10,
11].
CT Data and Preprocessing
We obtained scans of this phantom with a 64-MDCT scanner (Brilliance 64,
Philips Health care) with the following parameters: 120 kV; collimation, 64;
pitch, 1.2. Effective tube charge settings were used in steps of the square
root of two from the lowest tube charge value available: 10, 14, 20, 28, 40,
56, and 80 mAs. In this way, the noise in the images decreased by a constant
factor between successive values. For the first phase of the study, scans of
the phantom filled with the four iodine mixtures and air were obtained with
tube charge values increasing from 10 mAs until the value was reached at which
all polyps were visible for all observers. Because we expected that the task
for the observers would be more difficult for the second phase of the study,
higher tube charge values (20–80 mAs) were used in this phase. In the
second phase, only the two lowest con centrations were used. For each contrast
level, a scan at 300 mAs was made for reference purposes. All images were
reconstructed at a slice thickness of 0.9 mm and an increment of 0.45 mm with
reconstruction filter C (sharper standard filter). The pixel size was 0.6
mm.
Measurement of Contrast and Noise
Because it is difficult to exactly obtain the intended contrast levels, for
each intended contrast level we measured the actual contrast level, that is,
the difference in attenuation between the wall of the colonic phantom and its
content. We also measured the SD as a measure of noise in the images at the
location of the polyps. This measurement of noise also was performed at the
same position in a cross section of the drum that contained water only. This
approach made it possible to determine the influence of the tagged material on
noise. The measuring procedure is described in Appendix 1.
Observer Studies
Scans of the phantom were used to construct virtual colons for the observer
studies. These colons consisted of a large number of rings separated by thin
PMMA rings mimicking haustral folds. Before images of the virtual colons were
presented to the observers, a research fellow assessed the visibility of the
polyps for all contrast levels at the lowest tube charge setting (10 mAs). If
the polyps could absolutely not be missed, the particular contrast level was
not further evaluated by the observers. This measure was taken because a
visibility threshold cannot be determined if all polyps are highly visible.
For all other contrast levels, the visibility of polyps was determined in the
first phase of the study.
Three observers, an experienced abdominal radiologist, a resident in
radiology, and a research fellow reviewed all images. All observers had read
more than 250 colonoscopic ally verified CTC examinations before this study.
The observers reviewed the images using a 2D review method (ViewForum, Philips
Healthcare) and indicated the location of the polyp with a cursor. All
annotations were digitally stored. Because window and level settings can
influence the conspicuity of submerged polyps
[12], the window width and
level were preset by two researchers at a setting deemed optimal for polyp
detection at the given contrast level. These settings varied from 1,000/200 HU
for contrast of 300 HU to 2,400/550 HU for contrast of 1,000 HU. The
observers, however, were free to adjust the win dow setting to their
preference. This procedure is in accordance with that used in the clinical
evaluation of CTC scans.
First Phase of Study
In the first phase of the study, we determined the visibility of the polyps
in an eight-alternative forced-choice paradigm. In this paradigm a polyp
always was present in each ring at one of eight possible locations (Fig.
3A,
3B,
3C,
3D,
3E,
3F,
3G,
3H), and the observers had to
choose one of the eight locations. With this paradigm we could determine the
visibility of polyps in a time-efficient way.
For every contrast level, a virtual colon was composed of 80 rings. Each
virtual colon contained rings from scans made at five tube charge levels (10,
14, 20, 28, and 40 mAs), so each observer had to locate 16 polyps at each
combination of tube charge and contrast. Each ring was rotated so that the
polyp was present in one of eight possible clock-face positions on the phantom
wall. The choice of location was made at random (Fig.
3A,
3B,
3C,
3D,
3E,
3F,
3G,
3H).
After the reading by the three observers, a researcher not involved in
reading the data assessed the correctness of the annotations. On the basis of
these annotations, the number and average percentages of correct responses
were determined for every combination of tube charge level and contrast level.
Psychometric curves with constant steepness were fitted to this average
percentage of correct responses as a function of tube charge
[13]. For the fitting we
applied a maximum likelihood procedure
[14] for every contrast level.
At each contrast level we determined the tube charge for 90% correct responses
and the SD of this estimate
[15].
We also determined whether the number of correctly detected polyps
increased significantly with the density of the contrast material. We counted
the correctly detected polyps for each virtual colon for each observer and for
the three observers combined. A chi-square test (SPSS 15.0, SPSS) was used for
the comparison of proportions in two independent groups.
Contrast-to-Noise Ratio
The conspicuity of a polyp of a certain size depends both on the contrast
of the polyp to its surroundings and on the noise in the image, quantified by
the contrast-to-noise ratio (CNR)
[16,
17]. For different contrast
levels, the conspicuity at the tube charge for 90% correct responses is by
definition the same. The CNRs at these tube charge levels should therefore
also be equal. We checked whether this was indeed the case to cross-validate
our results.
The SD of the attenuation is used as a measure of noise. The CNR at 90%
correct responses was determined by dividing each contrast value by the SD for
that contrast value at the tube charge for 90% correct responses. The SD can
be considered inversely proportional to the square root of the tube charge
[18]. Therefore, the SD at the
tube charge for 90% correct responses can be derived from the SD in the 300
mAs reference scan as follows:
where mAs@90% is the tube charge value for 90% correct responses.
Second Phase of Study
Although in an eight-alternative forced-choice design, the tube charge
value for 90% correct responses can be determined efficiently, no clinically
applicable measures such as sen sitivity and number of false-positive findings
can be determined. In the second phase of the study, these performance values
were measured, and the threshold tube charge value for a sensitivity of at
least 90% was determined.
The second phase was performed at the two lowest contrast levels, 300 and
480 HU, and scans were obtained at five tube charge levels between 20 and 80
mAs. For each contrast and tube charge level, three virtual colons were
composed, each consisting of 120 rings
(Fig. 4). Five percent (18 of
360) of these rings contained a polyp. Rings containing polyps were randomly
distributed among the three colons. All rings were randomly rotated. Thus the
virtual colons in the second phase of the study differed in three important
aspects from those of the first phase: The prevalence of polyps was only 5%;
in each ring the polyps were situated at any angle along the phantom wall; and
colons were composed of rings scanned at the same tube charge levels.
View larger version (34K):
[in this window]
[in a new window]
[as a PowerPoint slide]
|
Fig. 4 —Coronal section of part of virtual colon (24 of 120 rings)
composed of scans at 80 mAs (width, 1,000 HU; level, 200 HU) for second phase
of study. Contrast between colon and its contents is 300 HU. Arrow indicates
6-mm sessile polyp.
|
|
The same three observers as in the first phase evaluated the virtual colons
starting at the level nearest the tube charge value for 90% cor rect responses
determined in the first phase. The tube charge was increased until the level
was reached at which the average per-polyp sensitivity of the observers was
greater than 90%.
The observers knew that the polyps, if present, were of the same size and
shape as in the first phase and were in random locations. The observers did
not, however, know the polyp prevalence. After the reading, a researcher not
involved in interpretation of the data assessed the correctness of the scores
by comparing the true polyp locations with the annotations of the observers.
For each contrast level, per-polyp sensitivity and the number of
false-positive findings were determined for every tube charge value used. The
tube charge value for 90% sensitivity was deter mined by linear
interpolation.
Also in the second phase, we determined whether the number of correctly
detected polyps significantly increased with the density of the tagging
material (300 and 480 HU). In this case, we compared the total numbers of
correctly detected polyps at 28 and 40 mAs for which data were available for
both densities. This procedure was performed for each observer and for the
three observers combined. The statistical procedure was the same as that used
in the first phase of the study.
Ratio of Thresholds of the First and Second Phases of the Study
The eight-alternative forced-choice task of the first phase of the study
was relatively easy. The second phase of the study was performed to obtain
results closer to the reality of clinical practice, but the task in this phase
was more difficult for the observers. The ratio of the tube charge value for
90% sensitivity (second phase) to the tube charge value for 90% correct
responses (first phase) was determined as an indicator of the relative
difficulty of detection of polyps in the second versus the first phase of the
study. This ratio can be used to translate the thresholds to extrapolate the
results of the first phase of the study to those of the more clinically
orientated second phase. The significance of this ratio is elucidated in the
Discussion section.
Results
Measurement of Contrast and Noise
The actual contrast values were close to the intended values. The actual
values of contrast and SD are listed in
Table 1.
Table 1 shows that the SD,
which is a measure of the noise at the location of the polyps, increases as
the contrast level increases. For the highest contrast level (1,040 HU), the
noise is approximately twice as high as for a phantom filled with air (SD, 70
and 33 HU, respectively).
Observer Study
The polyps in the air-filled colon were clearly visible at 10 mAs
(Fig. 5). Consequently, this
contrast level was not further evaluated.
View larger version (162K):
[in this window]
[in a new window]
[as a PowerPoint slide]
|
Fig. 5 —Polymethyl methacrylate colon phantom in center of
34-cm-diameter water-filled drum. CT scan (10 mAs; width, 2,000 HU; level, 0
HU) shows 6-mm polyp (arrow) surrounded by air. Polyp was considered
visible beyond doubt.
|
|
First Phase of Study
With increasing tube charge levels, the number of errors decreased. At
contrast levels of 300, 480, and 790 HU, the observers correctly located all
polyps at, respectively, 40, 28, and 20 mAs. The tube charges for 90% correct
responses for the three observers combined were 24.0, 16.3, and 6.2 mAs for
these three contrast levels (Fig.
6). At 790 HU, nearly all polyps were located correctly. The
fitting procedure remained valid under these circumstances
[14,
15], although the precision of
the estimate was impaired. This phenomenon was evidenced by the much higher SD
of the estimated threshold (Table
2). At 1,040 HU, all polyps were correctly located at all tube
charge levels. Therefore, the tube charge value for 90% correct responses
could not be determined for this contrast level.
For the 300, 480, and 790 HU contrast levels, the numbers of correctly
detected polyps in each virtual colon (i.e., for the five tube charge levels
combined) are listed in Table
3. Nearly all differences between contrast levels were
significant.
The CNRs at the tube charge value for 90% correct responses varied between
1.7 and 1.9, with an average of 1.8 (Table
2).
Second Phase of Study
Sensitivities and numbers of false-positive findings by the three observers
combined at contrast levels of 300 and 480 HU are shown in
Figure 7. Sensitivity
increased with tube charge level and with the density of the contrast
material. The sensitivity at a contrast of 300 HU exceeded 90% at 80 mAs and
at a contrast of 480 HU exceeded 90% at 40 mAs. Using linear interpolation, we
obtained tube charges for 90% sensitivity of 70 mAs and 35 mAs for 300 HU and
480 HU, respectively. At these tube charge levels, each observer had a maximum
of one false-positive finding. For 300 and 480 HU, the total numbers of
correctly detected polyps per observer at 28 and 40 mAs are listed in
Table 4. All observers detected
more polyps in the colon filled with the densest contrast material. This
difference was statistically significant for two of the three observers and
for all observers combined.
Ratio
The ratios of the tube charge for 90% sensitivity (second phase of the
study) and the tube charge for 90% correct responses (first phase of the
study) were 2.9 for 300 HU and 2.2 for 480 HU, with an average value of
2.5.
Discussion
The results of this CTC study show that the minimal radiation dose needed
to visualize a 6-mm polyp increases considerably if the contrast between the
polyp and its surroundings is reduced. Although all polyps in air were well
visible at 10 mAs, approximately 70 mAs was needed to achieve 90% sensitivity
at a contrast level of 300 HU.
We determined the minimal tube charge level to visualize a clinically
relevant polyp in two study phases. The first phase was performed to obtain
the threshold visibility in a simple and time-efficient way. It showed that if
contrast was increased from 300 to 790 HU, the tube charge for 90% correct
responses decreased from 24 to 6.2 mAs. The CNRs at these tube charge levels
were nearly constant. This finding can be expected because the detectability
of structures with relatively low contrast, such as polyps, is primarily
limited by noise on the images
[18]. With a 1.8 mean CNR at
the tube charge for 90% correct responses, the tube charge level for 90%
correct responses at the highest contrast (1,040 HU) and for air can be
estimated to be 4 and 1 mAs, respectively.
We realized that extrapolation to a clinical setting is a problem with
phantom studies. Therefore, we performed the second phase of the study to
obtain results closer to the reality of clinical practice. Important points of
the setup of the second phase are that only a few polyps were present, at
uncertain locations, and that measures of practical clinical relevance could
be determined. We expected that in this setup, the task for the observers in
the second phase of the study would be more difficult than in the first phase.
Our expectation turned out to be correct. In the second phase of the study, a
sensitivity of 90% at 300 and 480 HU was obtained at, respectively, 70 and 35
mAs, with less than one false-positive finding per observer per tube charge
level. Thus, this sensitivity was obtained at tube charge levels approximately
2.5 times higher than the corresponding levels in the first phase of the
study.
This ratio depends on factors such as the number of locations in which a
polyp can be present (only eight for the first phase of the study and at a
large number of locations in the second phase), the certainty or lack of
certainty that a polyp is present, and the possibility of choice (forced in
the first phase of the study, free in the second phase). Thus this ratio
reflects the relative difficulty of the tasks in the two phases, and the
ratios can therefore be expected to be similar for each contrast level.
The independence of contrast level was also found by Burgess and
Ghandeharian [19], who
compared the percentage of correct responses in a two-alternative
forced-choice study with those of alternative forced-choice studies with up to
1,800 possible locations. Therefore, we could have estimated this ratio using
only one of the contrast levels in the second phase of the study. For cross
check and to improve accuracy and precision, we decided to use two contrast
levels. We could have performed the second phase of the study with a third
contrast level (790 HU). However, we did not do so since nearly all polyps
were visible at this contrast level in the first phase of the study. As a
consequence, the precision of the threshold for the first phase was rather
poor (Table 3). Using the ratio
of 2.5, we can estimate that for the second phase of the study, the tube
charges for 90% sensitivity should be on the order of 15, 10, and 3 mAs for
contrast of 790 and 1,040 HU and for air.
Our results are for data from all three observers combined. Both observer
studies showed interobserver variability. This finding is evident from the
error bars in Figures 6 and
7. The results in Tables
3 and
4 show that systematic
differences were present between observers. Yet the general trend is confirmed
by the results for the individual observers, which consistently showed an
increase in polyp detectability with an increase in the density of the tagging
material. In most cases this increase was significant. We conclude that
although individual thresholds may vary, the influence of tagged material on
visibility is important.
A number of phantom studies and simulation studies have been conducted to
determine optimal scan parameters for CTC
[20–27].
In all of those studies, the visibility of polyps was investigated in air
only. Because of differences in study design and scan parameters, a comparison
of these studies is difficult. The main finding is that all polyps 6 mm in
diameter or larger are detected, even at the lowest tube charge levels (5 or
10 mAs). These studies are thus in line with our findings on polyps in
air.
For this study we used the range of contrast levels measured in tagged CTC
examinations at our hospital. Polyp detectability increases with the density
of the tagging material. Aiming at even higher densities than used in this
study may seem advantageous. It appears, however, that as contrast level
increases, image noise increases as well
(Table 1). This phenomenon is a
consequence of the increased attenuation of x-rays in dense material. The
increase in noise, also noted by Zalis et al.
[28], counteracts the improved
visibility owing to the higher contrast and may eventually lead to streak
artifacts in the directions of the highest attenuation of the x-rays. For
these reasons, striving for higher densities of tagging material than the ones
used in this study may be of limited value.
Many factors contribute to the quality of CT images. In this study we
varied the density of the contrast agent and the tube charge level. Other
image-quality factors such as tube voltage, pitch, and slice thickness were
left constant and were not subject to study. The conspicuity of polyps is also
considerably influenced by the size of the patient. In this study we used a
water-filled drum with a diameter of 34 cm, mimicking a rather corpulent
patient. If we had used a smaller human phantom, the noise would have been
lower, and therefore the same visibility of the polyps would have been
obtained with lower tube charge values. In state-of-the-art scanners, dose
modulation can be used in which the tube current can be automatically adapted
to the size of a patient. The results of this study can be used to determine
the reference setting for dose modulation.
With regard to ability to generalize our results to other CT scanners, it
is important to realize that CT scanners differ in effective dose and image
quality when scans are obtained with identical parameter settings. These
differences have to be taken into account in translation of our threshold
values to other scanners. The conclusion of our study, however, applies to any
CT scanner.
The main factor in visibility of a polyp is its size. The size of a polyp
correlates with its malignant potential
[29–31].
Polyps measuring 6–9 mm in diameter, especially polyps 10 mm in diameter
and larger, are associated with increased risk of abnormal histologic
findings. Polyps smaller than 6 mm can be disregarded because of minimal risk.
This policy is endorsed by a CTC consensus proposal by Zalis et al.
[9], which defines a normal
colon as a colon without polyps larger than 5 mm, and a European Society of
Gastrointestinal and Abdominal Radiology consensus statement
[32] that a reasonable minimum
size for reported polyps is 5 or 6 mm, depending on local preference.
We found that the visibility of polyps is substantially reduced when the
polyp is submerged in tagged material with relatively low density. In
practice, however, the overall per-polyp sensitivity would be affected less
because not all segments are completely filled with tagged material, even if a
regimen of reduced bowel preparation is undertaken. Owing to acquisition of
prone and supine images, a polyp may not be surrounded by tagged material in
both scan positions. We believe, however, that one should aim to visualize all
relevant polyps and thus choose the scan parameters to reach this goal. Doing
so means that the dose of the scan has to be increased in comparison with that
of a scan in which the colon is completely cleansed. The higher dose level
also may influence the visibility of extracolonic findings.
This study had limitations. Although the setup in the second phase was
chosen to more closely resemble a clinical situation than did the first phase,
it still is an idealization of real life. Polyp detection in patients can be
more difficult than it was in the second phase of the study. Thus tube charge
values needed in practice to adequately visualize 6-mm polyps may be somewhat
higher than in the study.
A second limitation was that the colon phantom was made of PMMA. This
material has an attenuation of approximately 120 HU, slightly higher than the
40-HU attenuation of colonic wall and polyps in vivo
[33]. We reported polyp
conspicuity as a function of the contrast, that is, the difference between the
attenuation of polyps and that of the luminal content. Our results therefore
also apply to these contrast levels in vivo in that each contrast level
corresponds to slightly lower attenuation of the tagging material. Because the
noise at the location of the polyp depends on the attenuation of the tagging
material (Table 1), the noise
would also be somewhat lower. One can calculate from the data in
Table 1 that this difference is
slight (
4%) and that the main outcome of our study remains
unaffected.
A third limitation was that we considered detection of polyps only in a
colon completely filled with contrast material. The increase in noise due to
increased attenuation in the tagged material would be lower in a smaller lumen
or a lumen only partly filled with tagged material. Therefore, in these
situations, the visibility of the polyps for the same tube charge level would
be somewhat better.
We used a homogeneous iodine solution to mimic colonic content because it
completely covered the inner surface of the phantom. It is known, however,
that different types of preparation produce tagging with varying degrees of
heterogeneity [11]. Therefore,
a fourth limitation was that the influence of heterogeneity of fecal material
was not evaluated. Heterogeneity may have a negative effect on the visibility
of colonic lesions. With inhomogeneous tagging, it can be expected that the
tube charge values needed to adequately visualize a 6-mm polyp will have to be
somewhat increased.
This phantom study is, to our knowledge, the first to show the negative
influence of tagging on the detectability of polyps covered by tagged
material. For adequate visualization of these polyps, a considerably higher
tube charge level is needed than for polyps in a cleansed colon, especially
when the density of the tagging material is low.
View larger version (10K):
[in this window]
[in a new window]
[as a PowerPoint slide]
|
Fig. 8A —Measuring contrast level and noise. Graphs show mean
(A) and SD (B) of attenuation in regions of interest with shape
of concentric rings within lumen of phantom and within polymethyl methacrylate
border. Circles indicate measured values; lines, fitted values. Value of
contrast and SD at site of polyp was taken as difference between fitted values
of contrast and SD for ring with diameter of 47 mm (indicated by dashed
vertical line), on which each polyp is centered. SDs for regions of interest
at 48 and 52 mm are high because these regions of interest contain pixels of
both lumen and border. SD at 50 mm is higher than 100 HU and therefore not
shown.
|
|
View larger version (9K):
[in this window]
[in a new window]
[as a PowerPoint slide]
|
Fig. 8B —Measuring contrast level and noise. Graphs show mean
(A) and SD (B) of attenuation in regions of interest with shape
of concentric rings within lumen of phantom and within polymethyl methacrylate
border. Circles indicate measured values; lines, fitted values. Value of
contrast and SD at site of polyp was taken as difference between fitted values
of contrast and SD for ring with diameter of 47 mm (indicated by dashed
vertical line), on which each polyp is centered. SDs for regions of interest
at 48 and 52 mm are high because these regions of interest contain pixels of
both lumen and border. SD at 50 mm is higher than 100 HU and therefore not
shown.
|
|
References
- van Gelder RE, Birnie E, Florie J, et al. CT colonography and
colonoscopy: assessment of patient preference in a 5-week follow-up study.
Radiology 2004;233
: 328–337[Abstract/Free Full Text]
- Iannaccone R, Laghi A, Catalano C, et al. Computed tomographic
colonography without cathartic preparation for the detection of colorectal
polyps. Gastroenterology 2004;127
:1300
–1311[CrossRef][Medline]
- Lefere P, Gryspeerdt S, Marrannes J, Baekelandt M, Van Holsbeeck B.
CT colonography after fecal tagging with a reduced cathartic cleansing and a
reduced volume of barium. AJR 2005;184
:1836
–1842[Abstract/Free Full Text]
- Gluecker TM, Johnson CD, Harmsen WS, et al. Colorectal cancer
screening with CT colonography, colonoscopy, and double-contrast barium enema
examination: prospective assessment of patient perceptions and preferences.
Radiology 2003;227
: 378–384[Abstract/Free Full Text]
- Rex DK. Current colorectal cancer screening strategies: overview
and obstacles to implementation. Rev Gastroenterol
Disord 2002; 2[suppl
1]: S2–S11
- Weitzman ER, Zapka J, Estabrook B, Goins KV. Risk and reluctance:
understanding impediments to colorectal cancer screening. Prev
Med 2001; 32:502
–513[CrossRef][Medline]
- Dachman AH, Dawson DO, Lefere P, et al. Comparison of routine and
unprepped CT colonography augmented by low fiber diet and stool tagging: a
pilot study. Abdom Imaging 2007;32
: 96–104[CrossRef][Medline]
- Zalis ME, Barish MA, Choi JR, et al. CT colonography reporting and
data system: a consensus proposal. Radiology2005; 236:3
–9[Free Full Text]
- Johnson KT, Carston MJ, Wentz RJ, Manduca A, Anderson SM, Johnson
CD. Development of a cathartic-free colorectal cancer screening test using
virtual colonoscopy: a feasibility study. AJR2007; 188:145; [web]W29
–W36[Abstract/Free Full Text]
- Zalis ME, Perumpillichira JJ, Magee C, Kohlberg G, Hahn PF.
Tagging-based, electronically cleansed CT colonography: evaluation of patient
comfort and image readability. Radiology2006; 239:149
–159[Abstract/Free Full Text]
- Slater A, Taylor SA, Burling D, Gartner L, Scarth J, Halligan S.
Colonic polyps: effect of attenuation of tagged fluid and viewing window on
conspicuity and measurement: in vitro experiment with porcine colonic
specimen. Radiology 2006;240
: 101–109[Abstract/Free Full Text]
- Veldkamp WJ, Thijssen MA, Karssemeijer N. The value of scatter
removal by a grid in full field digital mammography. Med
Phys 2003; 30:1712
–1718[CrossRef][Medline]
- Harvey LO Jr. Efficient estimation of sensory thresholds with
ML-PEST. Spat Vis 1997;11
: 121–128[Medline]
- Coppens JE, Franssen L, van den Berg TJ. Reliability of the
compensation comparison method for measuring retinal stray light studied using
Monte-Carlo simulations. J Biomed Opt2006; 11:054010[CrossRef][Medline]
- Nakayama Y, Awai K, Funama Y, et al. Lower tube voltage reduces
contrast material and radiation doses on 16-MDCT aortography.
AJR 2006; 187:
1266; [web] W490–W497[Abstract/Free Full Text]
- Doyle P, Martin CJ, Gentle D. Application of contrast-to-noise
ratio in optimizing beam quality for digital chest radiography: comparison of
experimental measurements and theoretical simulations. Phys Med
Biol 2006; 51:2953
–2970[CrossRef][Medline]
- Kalender W. Computed tomography. Erlangen,
Germany: Publicis, 2005:98
–153
- Burgess AE, Ghandeharian H. Visual signal detection. Part 2.
Signal-location identification. J Opt Soc Am A1984; 1:906
–910[Medline]
- Embleton KV, Nicholson DA, Hufton AP, Jackson A. Optimization of
scanning parameters for multi-slice CT colonography: experiments with
synthetic and animal phantoms. Clin Radiol2003; 58:955
–963[CrossRef][Medline]
- Johnson KT, Johnson CD, Anderson SM, Bruesewitz MR, Mccollough CH.
CT colonography: determination of optimal CT technique using a novel colon
phantom. Abdom Imaging 2004;29
: 173–176[Medline]
- Ozgun A, Rollven E, Blomqvist L, Bremmer S, Odh R, Fransson A.
Polyp detection with MDCT: a phantom-based evaluation of the impact of dose
and spatial resolution. AJR 2005;184
:1181
–1188[Abstract/Free Full Text]
- Taylor SA, Halligan S, Bartram CI, et al. Multidetector row CT
colonography: effect of collimation, pitch, and orientation on polyp detection
in a human colectomy specimen. Radiology2003; 229:109
–118[Abstract/Free Full Text]
- Wessling J, Fischbach R, Meier N, et al. CT colonography: protocol
optimization with multidetector row CT—study in an anthropomorphic colon
phantom. Radiology 2003;228
: 753–759[Abstract/Free Full Text]
- Won HJ, Choi BI, Kim SH, Kim YI, Youn BJ, Han JK. Protocol
optimization of multidetector computed tomography colonography using pig
colonic phantoms. Invest Radiol 2005;40
: 27–32[Medline]
- van Gelder RE, Venema HW, Florie J, et al. CT colonography:
feasibility of substantial dose reduction—comparison of medium to very
low doses in identical patients. Radiology2004; 232:611
–620[Abstract/Free Full Text]
- Wessling J, Fischbach R, Borchert A, et al. Detection of colorectal
polyps: comparison of multidetector row CT and MR colonography in a colon
phantom. Radiology 2006;241
: 125–131[Abstract/Free Full Text]
- Zalis ME, Perumpillichira JJ, Kim JY, Del FC, Magee C, Hahn PF.
Polyp size at CT colonography after electronic subtraction cleansing in an
anthropomorphic colon phantom. Radiology2005; 236:118
–124[Abstract/Free Full Text]
- Atkin WS, Morson BC, Cuzick J. Long-term risk of colorectal cancer
after excision of rectosigmoid adenomas. N Engl J Med1992; 326:658
–662[Abstract]
- O'Brien MJ, Winawer SJ, Zauber AG, et al. The National Polyp Study:
patient and polyp characteristics associated with high-grade dysplasia in
colorectal adenomas. Gastroenterology1990; 98:371
–379[Medline]
- Simons BD, Morrison AS, Lev R, Verhoek-Oftedahl W. Relationship of
polyps to cancer of the large intestine. J Natl Cancer
Inst 1992; 84:962
–966[Abstract/Free Full Text]
- Taylor SA, Laghi A, Lefere P, Halligan S, Stoker J. European
Society of Gastrointestinal and Abdominal Radiology (ESGAR): consensus
statement on CT colonography. Eur Radiol2007; 17:575
–579[CrossRef][Medline]
- Neri E, Vagli P, Picchietti S, et al. CT colonography: contrast
enhancement of benign and malignant colorectal lesions versus fecal residuals.
Abdom Imaging 2005;30
: 694–697[CrossRef][Medline]
CiteULike 
Complore 
Connotea 
Del.icio.us 
Digg 
Reddit 
Technorati What's this?
Top
Abstract
Introduction
= 400 HU, 
= 790 HU.
= number of false-positive findings at 480 HU, 
= number of
false-positive findings at 300 HU.